From 0089e59187e71ccee22a24e908a95b50408c4ca0 Mon Sep 17 00:00:00 2001
From: ruk0sh <43880105+ruk0sh@users.noreply.github.com>
Date: Fri, 15 Apr 2022 05:28:38 -0700
Subject: [PATCH 01/17] Adapt raw files preprocessing for new loaders, add
emo_speaker_file preparation, add deps for Excel parsing to conda env
---
environment.yml | 2 ++
russian_preprocessing/steps.sh | 7 ++---
src/preprocessing/prep_files_russian.py | 42 +++++++++++++++++++------
3 files changed, 36 insertions(+), 15 deletions(-)
diff --git a/environment.yml b/environment.yml
index 3446865..0379f6c 100644
--- a/environment.yml
+++ b/environment.yml
@@ -18,6 +18,7 @@ dependencies:
- matplotlib
- montreal-forced-aligner=2.0.0b4
- numpy
+ - openpyxl
- pip
- python=3.8
- pytorch=1.10.0
@@ -29,6 +30,7 @@ dependencies:
- unidecode
- unzip
- uvicorn
+ - xlrd
- pip:
- tgt
\ No newline at end of file
diff --git a/russian_preprocessing/steps.sh b/russian_preprocessing/steps.sh
index 4ee2f2e..08b684f 100644
--- a/russian_preprocessing/steps.sh
+++ b/russian_preprocessing/steps.sh
@@ -3,15 +3,12 @@ conda env create -n emotts -f russian_preprocessing/environment.yaml
conda activate emotts
conda config --set ssl_verify no
export RUSSIAN_DATASET_PATH=/media/diskB/ruslan_a/data/datasets/EMO/russian
-export OUTPUT_DIR=$RUSSIAN_DATASET_PATH/processed
-export OUTPUT_DIR=$RUSSIAN_DATASET_PATH/processed_mix
+export OUTPUT_DIR=$RUSSIAN_DATASET_PATH/processed_v2
export MFA_PREMADE=/media/diskB/ruslan_a/data/datasets/emo_rus_Olga_v2_processed/mfa_espeak_grids
# 16164it [03:36, 74.61it/s]
-# 25544it [02:45, 154.48it/s] mix
echo -e "\n1) Prep raw files"
-# python src/preprocessing/prep_files_russian.py --dataset-dir $RUSSIAN_DATASET_PATH/original --text-output-dir $OUTPUT_DIR/text/raw --audio-output-dir $OUTPUT_DIR/audio/raw
-python src/preprocessing/prep_files_russian.py --dataset-dir $RUSSIAN_DATASET_PATH/original_and_neutral --text-output-dir $OUTPUT_DIR/text/raw --audio-output-dir $OUTPUT_DIR/audio/raw
+python src/preprocessing/prep_files_russian.py --dataset-dir $RUSSIAN_DATASET_PATH/original --text-output-dir $OUTPUT_DIR/text/raw --audio-output-dir $OUTPUT_DIR/audio/raw --meta-output-dir $OUTPUT_DIR/meta
# ~1.5-2.0 hours
echo -e "\n2) Pausation cutting with VAD"
diff --git a/src/preprocessing/prep_files_russian.py b/src/preprocessing/prep_files_russian.py
index d47900d..04cb3e9 100644
--- a/src/preprocessing/prep_files_russian.py
+++ b/src/preprocessing/prep_files_russian.py
@@ -1,5 +1,7 @@
#!/usr/bin/env python
+import json
import shutil
+from collections import defaultdict
from pathlib import Path
from typing import List
@@ -8,19 +10,22 @@
from tqdm import tqdm
-def process_audio(audio_path: Path, audio_output_dir: Path) -> None:
- speaker = audio_path.parent.name.replace("_", "-")
+def process_audio(audio_path: Path, audio_output_dir: Path, emo_speaker: defaultdict) -> None:
+ old_dir = audio_path.parent.name.replace("_", "-")
+ speaker = "olga"
emotion = audio_path.parent.parent.name
new_dir = audio_output_dir / speaker
new_dir.mkdir(parents=True, exist_ok=True)
- new_filename = f"{speaker}_{emotion}_{audio_path.name}"
+ new_filename = f"{old_dir}-{emotion}_{audio_path.name}"
new_audio_path = new_dir / new_filename
+ emo_speaker[emotion][speaker].add(new_audio_path.stem)
shutil.copy(audio_path, new_audio_path)
-def process_annotation(annot_path: Path, text_output_dir: Path) -> None:
+def process_annotation(annot_path: Path, text_output_dir: Path, emo_speaker: defaultdict) -> None:
text_ext = "txt"
- speaker = annot_path.parent.name.replace("_", "-")
+ old_dir = annot_path.parent.name.replace("_", "-")
+ speaker = "olga"
emotion = annot_path.parent.parent.name
df: pd.DataFrame = pd.read_excel(annot_path)
df = df.iloc[:, :3]
@@ -32,11 +37,12 @@ def process_annotation(annot_path: Path, text_output_dir: Path) -> None:
except KeyError:
print(row)
raise
- new_filename = f"{speaker}_{emotion}_{filename}.{text_ext}"
+ new_filename = f"{old_dir}-{emotion}_{filename}.{text_ext}"
new_dir = text_output_dir / speaker
new_dir.mkdir(parents=True, exist_ok=True)
new_filepath = new_dir / new_filename
- with open(new_filepath, "w") as text_output_file:
+ emo_speaker[emotion][speaker].add(new_filepath.stem)
+ with open(new_filepath, "w", encoding="utf8") as text_output_file:
text_output_file.write(content)
@@ -56,6 +62,12 @@ def process_annotation(annot_path: Path, text_output_dir: Path) -> None:
default="wavs",
help="Directory for rearranged audio files.",
)
+@click.option(
+ "--meta-output-dir",
+ type=Path,
+ default="meta",
+ help="Directory for newly created metadata files.",
+)
@click.option(
"--log-path",
type=Path,
@@ -74,6 +86,7 @@ def main(
dataset_dir: Path,
text_output_dir: Path,
audio_output_dir: Path,
+ meta_output_dir: Path,
log_path: Path,
audio_ext: str,
annot_ext: List[str],
@@ -81,23 +94,32 @@ def main(
text_output_dir.mkdir(exist_ok=True, parents=True)
audio_output_dir.mkdir(exist_ok=True, parents=True)
+ meta_output_dir.mkdir(exist_ok=True, parents=True)
log_path.parent.mkdir(exist_ok=True, parents=True)
log_path.unlink(missing_ok=True)
- for path in tqdm(dataset_dir.rglob("*")):
+ emo_speaker_json = defaultdict(lambda: defaultdict(set))
+ for path in tqdm(list(dataset_dir.rglob("*"))):
# If audio, get speaker, get emotion and copy it with new name
if path.suffix == f".{audio_ext}":
- process_audio(path, audio_output_dir)
+ process_audio(path, audio_output_dir, emo_speaker_json)
# If annotation, parse it and rearrange texts
elif path.suffix[1:] in annot_ext:
- process_annotation(path, text_output_dir)
+ process_annotation(path, text_output_dir, emo_speaker_json)
# else do nothing, log skipped path
else:
with open(log_path, "a") as logfile:
logfile.write(f"{path}\n")
continue
+ emo_speaker_json = dict(emo_speaker_json)
+ for emotion in emo_speaker_json.keys():
+ for speaker in emo_speaker_json[emotion].keys():
+ emo_speaker_json[emotion][speaker] = list(emo_speaker_json[emotion][speaker])
+ with open(meta_output_dir / "emo_speaker_file.json", "w", encoding="utf8") as f:
+ json.dump(emo_speaker_json, f)
+
if __name__ == "__main__":
main()
From 6c84a209ca8c361b409ee472c5e4f54e3f7b45ac Mon Sep 17 00:00:00 2001
From: ruk0sh <43880105+ruk0sh@users.noreply.github.com>
Date: Fri, 15 Apr 2022 06:09:40 -0700
Subject: [PATCH 02/17] Add speaker directory creation
---
russian_preprocessing/steps.sh | 8 +++-----
src/preprocessing/wav_to_mel.py | 4 +++-
2 files changed, 6 insertions(+), 6 deletions(-)
diff --git a/russian_preprocessing/steps.sh b/russian_preprocessing/steps.sh
index 08b684f..f6a3d4d 100644
--- a/russian_preprocessing/steps.sh
+++ b/russian_preprocessing/steps.sh
@@ -10,12 +10,11 @@ export MFA_PREMADE=/media/diskB/ruslan_a/data/datasets/emo_rus_Olga_v2_processed
echo -e "\n1) Prep raw files"
python src/preprocessing/prep_files_russian.py --dataset-dir $RUSSIAN_DATASET_PATH/original --text-output-dir $OUTPUT_DIR/text/raw --audio-output-dir $OUTPUT_DIR/audio/raw --meta-output-dir $OUTPUT_DIR/meta
-# ~1.5-2.0 hours
-echo -e "\n2) Pausation cutting with VAD"
-python src/preprocessing/pausation_cutting.py --input-dir $OUTPUT_DIR/audio/raw --output-dir $OUTPUT_DIR/audio/no_pause --target-sr 96000 --audio-ext wav
+# ~1.5-2.0 hours (Deprecated)
+# echo -e "\n2) Pausation cutting with VAD"
+# python src/preprocessing/pausation_cutting.py --input-dir $OUTPUT_DIR/audio/raw --output-dir $OUTPUT_DIR/audio/no_pause --target-sr 96000 --audio-ext wav
# 16071/16071 [02:50<00:00, 94.26it/s]
-# 25429/25429 [04:18<00:00, 98.26it/s] mix
echo -e "\n3) Resampling and Converting audio to 1-channel"
# python src/preprocessing/resampling.py --input-dir $OUTPUT_DIR/audio/no_pause --output-dir $OUTPUT_DIR/audio/resampled --resample-rate 22050 --audio-ext wav
python src/preprocessing/resampling.py --input-dir $OUTPUT_DIR/audio/raw --output-dir $OUTPUT_DIR/audio/resampled --resample-rate 22050 --audio-ext wav
@@ -25,7 +24,6 @@ conda deactivate
conda activate emotts
# 16071/16071 [13:09<00:00, 20.37it/s]
-# 25429/25429 [13:11<00:00, 32.15it/s] mix
echo -e "\n4) Audio to Mel"
python src/preprocessing/wav_to_mel.py --input-dir $OUTPUT_DIR/audio/resampled --output-dir $OUTPUT_DIR/mels --audio-ext wav
diff --git a/src/preprocessing/wav_to_mel.py b/src/preprocessing/wav_to_mel.py
index 604039c..20ffe2b 100644
--- a/src/preprocessing/wav_to_mel.py
+++ b/src/preprocessing/wav_to_mel.py
@@ -66,7 +66,9 @@ def main(input_dir: Path, output_dir: Path, audio_ext: str) -> None:
print("Transforming audio to mel...")
for filepath in tqdm(filepath_list):
- new_path = output_dir / filepath.stem
+ speaker = filepath.parent.stem
+ new_path = output_dir / speaker / filepath.stem
+ new_path.parent.mkdir(exist_ok=True, parents=True)
wave_tensor, _ = torchaudio.load(filepath)
assert wave_tensor.shape[0] == 1, "Audio has more than 1 channel"
From aec36e051284986df08c5bf660c54f91511bd5b1 Mon Sep 17 00:00:00 2001
From: ruk0sh <43880105+ruk0sh@users.noreply.github.com>
Date: Fri, 15 Apr 2022 07:15:57 -0700
Subject: [PATCH 03/17] Fix encoding
---
src/preprocessing/text_normalization_russian.py | 13 +++++++------
1 file changed, 7 insertions(+), 6 deletions(-)
diff --git a/src/preprocessing/text_normalization_russian.py b/src/preprocessing/text_normalization_russian.py
index 946aa5f..591938b 100644
--- a/src/preprocessing/text_normalization_russian.py
+++ b/src/preprocessing/text_normalization_russian.py
@@ -27,12 +27,13 @@ def main(input_dir: Path, output_dir: Path) -> None:
new_dir.mkdir(exist_ok=True)
new_file = new_dir / filepath.name
- with open(filepath, "r") as fin, open(new_file, "w") as fout:
- content = fin.read()
- normalized_content = normalizer.norm_text(content)
- normalized_content = lowercase(normalized_content)
- normalized_content = collapse_whitespace(normalized_content)
- fout.write(normalized_content)
+ with open(filepath, "r", encoding="utf8") as fin:
+ with open(new_file, "w", encoding="utf8") as fout:
+ content = fin.read()
+ normalized_content = normalizer.norm_text(content)
+ normalized_content = lowercase(normalized_content)
+ normalized_content = collapse_whitespace(normalized_content)
+ fout.write(normalized_content)
print("Finished successfully.")
print(f"Processed files are located at {output_dir}")
From bfce4462e670b5d869ed478720d6fa48f76f44da Mon Sep 17 00:00:00 2001
From: ruk0sh <43880105+ruk0sh@users.noreply.github.com>
Date: Mon, 18 Apr 2022 04:22:03 -0700
Subject: [PATCH 04/17] Add neutral raw russian data preprocessing
---
russian_preprocessing/steps.sh | 38 +++---
.../prep_files_russian_neutral.py | 115 ++++++++++++++++++
2 files changed, 137 insertions(+), 16 deletions(-)
create mode 100644 src/preprocessing/prep_files_russian_neutral.py
diff --git a/russian_preprocessing/steps.sh b/russian_preprocessing/steps.sh
index f6a3d4d..f10bf8c 100644
--- a/russian_preprocessing/steps.sh
+++ b/russian_preprocessing/steps.sh
@@ -2,22 +2,26 @@ echo -e "\n0) Setup env"
conda env create -n emotts -f russian_preprocessing/environment.yaml
conda activate emotts
conda config --set ssl_verify no
-export RUSSIAN_DATASET_PATH=/media/diskB/ruslan_a/data/datasets/EMO/russian
-export OUTPUT_DIR=$RUSSIAN_DATASET_PATH/processed_v2
+export EMO_DATASET_DIR=/media/diskB/ruslan_a/data/datasets/EMO/russian
+export NEUTRAL_DATASET_DIR=/media/diskB/ruslan_a/data/datasets/NEUTRAL/russian
+export EMO_OUTPUT_DIR=$EMO_DATASET_DIR/processed_v2
+export NEUTRAL_OUTPUT_DIR=$NEUTRAL_DATASET_DIR/processed_v2
export MFA_PREMADE=/media/diskB/ruslan_a/data/datasets/emo_rus_Olga_v2_processed/mfa_espeak_grids
# 16164it [03:36, 74.61it/s]
echo -e "\n1) Prep raw files"
-python src/preprocessing/prep_files_russian.py --dataset-dir $RUSSIAN_DATASET_PATH/original --text-output-dir $OUTPUT_DIR/text/raw --audio-output-dir $OUTPUT_DIR/audio/raw --meta-output-dir $OUTPUT_DIR/meta
+python src/preprocessing/prep_files_russian.py --dataset-dir $EMO_DATASET_DIR/original --text-output-dir $EMO_OUTPUT_DIR/text/raw --audio-output-dir $EMO_OUTPUT_DIR/audio/raw --meta-output-dir $EMO_OUTPUT_DIR/meta
+python src/preprocessing/prep_files_russian_neutral.py --dataset-dir $NEUTRAL_DATASET_DIR/original --text-output-dir $NEUTRAL_OUTPUT_DIR/text/raw --audio-output-dir $NEUTRAL_OUTPUT_DIR/audio/raw --meta-output-dir $NEUTRAL_OUTPUT_DIR/meta
# ~1.5-2.0 hours (Deprecated)
# echo -e "\n2) Pausation cutting with VAD"
-# python src/preprocessing/pausation_cutting.py --input-dir $OUTPUT_DIR/audio/raw --output-dir $OUTPUT_DIR/audio/no_pause --target-sr 96000 --audio-ext wav
+# python src/preprocessing/pausation_cutting.py --input-dir $EMO_OUTPUT_DIR/audio/raw --output-dir $EMO_OUTPUT_DIR/audio/no_pause --target-sr 96000 --audio-ext wav
# 16071/16071 [02:50<00:00, 94.26it/s]
echo -e "\n3) Resampling and Converting audio to 1-channel"
-# python src/preprocessing/resampling.py --input-dir $OUTPUT_DIR/audio/no_pause --output-dir $OUTPUT_DIR/audio/resampled --resample-rate 22050 --audio-ext wav
-python src/preprocessing/resampling.py --input-dir $OUTPUT_DIR/audio/raw --output-dir $OUTPUT_DIR/audio/resampled --resample-rate 22050 --audio-ext wav
+# python src/preprocessing/resampling.py --input-dir $EMO_OUTPUT_DIR/audio/no_pause --output-dir $EMO_OUTPUT_DIR/audio/resampled --resample-rate 22050 --audio-ext wav
+python src/preprocessing/resampling.py --input-dir $EMO_OUTPUT_DIR/audio/raw --output-dir $EMO_OUTPUT_DIR/audio/resampled --resample-rate 22050 --audio-ext wav
+python src/preprocessing/resampling.py --input-dir $NEUTRAL_OUTPUT_DIR/audio/raw --output-dir $NEUTRAL_OUTPUT_DIR/audio/resampled --resample-rate 22050 --audio-ext wav
conda env config vars set LD_LIBRARY_PATH=$CONDA_PREFIX/lib # link to libopenblas
conda deactivate
@@ -25,14 +29,16 @@ conda activate emotts
# 16071/16071 [13:09<00:00, 20.37it/s]
echo -e "\n4) Audio to Mel"
-python src/preprocessing/wav_to_mel.py --input-dir $OUTPUT_DIR/audio/resampled --output-dir $OUTPUT_DIR/mels --audio-ext wav
+python src/preprocessing/wav_to_mel.py --input-dir $EMO_OUTPUT_DIR/audio/resampled --output-dir $EMO_OUTPUT_DIR/mels --audio-ext wav
+python src/preprocessing/wav_to_mel.py --input-dir $NEUTRAL_OUTPUT_DIR/audio/resampled --output-dir $NEUTRAL_OUTPUT_DIR/mels --audio-ext wav
# 16069/16069 [31:38<00:00, 8.46it/s]
# 16962/16962 [12:22<00:00, 22.85it/s] mix
# Model is needed: https://github.com/snakers4/russian_stt_text_normalization/blob/master/jit_s2s.pt
# Put model to src/preprocessing/text/russian/
echo -e "\n5) Text normalization"
-python src/preprocessing/text_normalization_russian.py --input-dir $OUTPUT_DIR/text/raw --output-dir $OUTPUT_DIR/mfa_inputs
+python src/preprocessing/text_normalization_russian.py --input-dir $EMO_OUTPUT_DIR/text/raw --output-dir $EMO_OUTPUT_DIR/mfa_inputs
+python src/preprocessing/text_normalization_russian.py --input-dir $NEUTRAL_OUTPUT_DIR/text/raw --output-dir $NEUTRAL_OUTPUT_DIR/mfa_inputs
echo -e "\n6) MFA Alignment setup"
@@ -47,33 +53,33 @@ export RUS_ESPEAK_LEXICON=/media/diskB/ruslan_a/models/mfa/rus-espeak-mfa/rus-mf
# 16069/16069 [00:00<00:00, 21508.15it/s]
echo -e "\n6.1) Creating word list from dataset"
-python src/preprocessing/create_corpus.py --input-dir $OUTPUT_DIR/text/raw --output-path $OUTPUT_DIR/meta/words.txt
+python src/preprocessing/create_corpus.py --input-dir $EMO_OUTPUT_DIR/text/raw --output-path $EMO_OUTPUT_DIR/meta/words.txt
# 25849/25983 [01:19<00:00, 326.82it/s]
echo -e "\n6.2) Creating G2P lexicon from word list"
-mfa g2p -t mfa_tmp -j 32 --clean --overwrite models/g2p/russian_g2p.zip $OUTPUT_DIR/meta/words.txt models/mfa/russian_lexicon.txt
+mfa g2p -t mfa_tmp -j 32 --clean --overwrite models/g2p/russian_g2p.zip $EMO_OUTPUT_DIR/meta/words.txt models/mfa/russian_lexicon.txt
rm -rf mfa_tmp
# 42it [00:10, 4.01it/s]
# 58it [00:12, 4.54it/s] mix
echo -e "\n7) MFA Preprocessing"
-python src/preprocessing/mfa_preprocessing.py --input-dir $OUTPUT_DIR/audio/resampled --output-dir $OUTPUT_DIR/mfa_inputs
+python src/preprocessing/mfa_preprocessing.py --input-dir $EMO_OUTPUT_DIR/audio/resampled --output-dir $EMO_OUTPUT_DIR/mfa_inputs
# FINALLY, align phonemes and speech
# ~50 min
echo -e "\n8) MFA Alignment"
-echo $OUTPUT_DIR
-mfa align -t mfa_tmp --clean -j 32 $OUTPUT_DIR/mfa_inputs models/mfa/russian_lexicon.txt models/mfa/russian.zip $OUTPUT_DIR/mfa_outputs
+echo $EMO_OUTPUT_DIR
+mfa align -t mfa_tmp --clean -j 32 $EMO_OUTPUT_DIR/mfa_inputs models/mfa/russian_lexicon.txt models/mfa/russian.zip $EMO_OUTPUT_DIR/mfa_outputs
# for mix
-mfa align -t mfa_tmp --clean -j 32 $OUTPUT_DIR/mfa_inputs $RUS_ESPEAK_LEXICON $RUS_ESPEAK_ACOUSTIC_MODEL $OUTPUT_DIR/mfa_outputs
+mfa align -t mfa_tmp --clean -j 32 $EMO_OUTPUT_DIR/mfa_inputs $RUS_ESPEAK_LEXICON $RUS_ESPEAK_ACOUSTIC_MODEL $EMO_OUTPUT_DIR/mfa_outputs
rm -rf mfa_tmp
echo -e "\n9) MFA Postprocessing"
# Aggregate mels by speakers
# 16071it [00:00, 16767.84it/s]
# 25429it [00:01, 17870.70it/s] mix
-python src/preprocessing/mfa_postprocessing.py --input-dir $OUTPUT_DIR/mels
+python src/preprocessing/mfa_postprocessing.py --input-dir $EMO_OUTPUT_DIR/mels
# Only for MFA PREMADE case
echo -e "\n10) MFA Premade Preprocessing"
-python src/preprocessing/mfa_premade_preprocessing.py --input-dir $MFA_PREMADE --output-dir $OUTPUT_DIR/mfa_outputs
\ No newline at end of file
+python src/preprocessing/mfa_premade_preprocessing.py --input-dir $MFA_PREMADE --output-dir $EMO_OUTPUT_DIR/mfa_outputs
\ No newline at end of file
diff --git a/src/preprocessing/prep_files_russian_neutral.py b/src/preprocessing/prep_files_russian_neutral.py
new file mode 100644
index 0000000..0dab729
--- /dev/null
+++ b/src/preprocessing/prep_files_russian_neutral.py
@@ -0,0 +1,115 @@
+#!/usr/bin/env python
+import json
+import shutil
+from collections import defaultdict
+from pathlib import Path
+
+import click
+import pandas as pd
+from tqdm import tqdm
+
+
+def process_audio(audio_path: Path, audio_output_dir: Path, emo_speaker: defaultdict) -> None:
+ old_dir = audio_path.parent.name.replace("_", "-")
+ speaker = "olga"
+ emotion = "neutral"
+ new_dir = audio_output_dir / speaker
+ new_dir.mkdir(parents=True, exist_ok=True)
+ new_filename = f"{old_dir}-{emotion}_{audio_path.name}"
+ new_audio_path = new_dir / new_filename
+ emo_speaker[emotion][speaker].add(new_audio_path.stem)
+ shutil.copy(audio_path, new_audio_path)
+
+
+def process_metadata(metadata_path: Path, text_output_dir: Path, emo_speaker: defaultdict) -> None:
+ text_ext = "txt"
+ speaker = "olga"
+ emotion = "neutral"
+ df: pd.DataFrame = pd.read_csv(metadata_path, delimiter="|", header=None, names=["path", "original", "stressed"])
+ for (_, path, original, stressed) in df.itertuples():
+ old_dir, filename = path.split("/")
+ new_filename = f"{old_dir}-{emotion}_{filename}.{text_ext}"
+ new_dir = text_output_dir / speaker
+ new_dir.mkdir(parents=True, exist_ok=True)
+ new_filepath = new_dir / new_filename
+ emo_speaker[emotion][speaker].add(new_filepath.stem)
+ with open(new_filepath, "w", encoding="utf8") as text_output_file:
+ text_output_file.write(original)
+
+
+@click.command()
+@click.option(
+ "--dataset-dir", type=Path, help="Directory with original russian dataset"
+)
+@click.option(
+ "--text-output-dir",
+ type=Path,
+ default="texts",
+ help="Directory for text files extracted from annotations.",
+)
+@click.option(
+ "--audio-output-dir",
+ type=Path,
+ default="wavs",
+ help="Directory for rearranged audio files.",
+)
+@click.option(
+ "--meta-output-dir",
+ type=Path,
+ default="meta",
+ help="Directory for newly created metadata files.",
+)
+@click.option(
+ "--log-path",
+ type=Path,
+ default="logs/preprocessing/russian-skipped-paths.txt",
+ help="Path for logging list of skipped items.",
+)
+@click.option(
+ "--metadata-filename",
+ type=str,
+ default="metadata.csv",
+ help="Name of metadata file to search in dataset.",
+)
+@click.option("--audio-ext", type=str, default="wav", help="Extension of audio files.")
+def main(
+ dataset_dir: Path,
+ text_output_dir: Path,
+ audio_output_dir: Path,
+ meta_output_dir: Path,
+ log_path: Path,
+ audio_ext: str,
+ metadata_filename: str,
+) -> None:
+
+ text_output_dir.mkdir(exist_ok=True, parents=True)
+ audio_output_dir.mkdir(exist_ok=True, parents=True)
+ meta_output_dir.mkdir(exist_ok=True, parents=True)
+ log_path.parent.mkdir(exist_ok=True, parents=True)
+ log_path.unlink(missing_ok=True)
+
+ emo_speaker_json = defaultdict(lambda: defaultdict(set))
+
+ for path in tqdm(list(dataset_dir.rglob("*"))):
+ # If audio, get speaker, get emotion and copy it with new name
+ if path.suffix == f".{audio_ext}":
+ process_audio(path, audio_output_dir, emo_speaker_json)
+ # If metadata, parse it and rearrange texts
+ elif path.name == metadata_filename:
+ process_metadata(path, text_output_dir, emo_speaker_json)
+ # else do nothing, log skipped path
+ else:
+ with open(log_path, "a") as logfile:
+ logfile.write(f"{path}\n")
+ continue
+
+ emo_speaker_json = dict(emo_speaker_json)
+ for emotion in emo_speaker_json.keys():
+ for speaker in emo_speaker_json[emotion].keys():
+ emo_speaker_json[emotion][speaker] = list(emo_speaker_json[emotion][speaker])
+ with open(meta_output_dir / "neutral_speaker_file.json", "w", encoding="utf8") as f:
+ json.dump(emo_speaker_json, f)
+
+
+if __name__ == "__main__":
+ main()
From 01a9c49fcc04077f84a4e823c9053db553b1c773 Mon Sep 17 00:00:00 2001
From: ruk0sh <43880105+ruk0sh@users.noreply.github.com>
Date: Mon, 18 Apr 2022 05:42:26 -0700
Subject: [PATCH 05/17] Fix text files renaming pattern
---
src/preprocessing/prep_files_russian_neutral.py | 1 +
1 file changed, 1 insertion(+)
diff --git a/src/preprocessing/prep_files_russian_neutral.py b/src/preprocessing/prep_files_russian_neutral.py
index 0dab729..5a8a160 100644
--- a/src/preprocessing/prep_files_russian_neutral.py
+++ b/src/preprocessing/prep_files_russian_neutral.py
@@ -28,6 +28,7 @@ def process_metadata(metadata_path: Path, text_output_dir: Path, emo_speaker: de
df: pd.DataFrame = pd.read_csv(metadata_path, delimiter="|", header=None, names=["path", "original", "stressed"])
for (_, path, original, stressed) in df.itertuples():
old_dir, filename = path.split("/")
+ old_dir = old_dir.replace("_", "-")
new_filename = f"{old_dir}-{emotion}_{filename}.{text_ext}"
new_dir = text_output_dir / speaker
new_dir.mkdir(parents=True, exist_ok=True)
From 2c2369b23d5cd90e754189190a27345c133aae02 Mon Sep 17 00:00:00 2001
From: ruk0sh <43880105+ruk0sh@users.noreply.github.com>
Date: Tue, 19 Apr 2022 13:26:55 +0300
Subject: [PATCH 06/17] Implement espeak-based phonemizer prototype
---
environment.yml | 1 +
src/preprocessing/phonemizer.py | 80 +++++++++++++++++++++++++++++++++
2 files changed, 81 insertions(+)
create mode 100644 src/preprocessing/phonemizer.py
diff --git a/environment.yml b/environment.yml
index 0379f6c..aca4305 100644
--- a/environment.yml
+++ b/environment.yml
@@ -32,5 +32,6 @@ dependencies:
- uvicorn
- xlrd
- pip:
+ - phonemizer
- tgt
\ No newline at end of file
diff --git a/src/preprocessing/phonemizer.py b/src/preprocessing/phonemizer.py
new file mode 100644
index 0000000..445b9e5
--- /dev/null
+++ b/src/preprocessing/phonemizer.py
@@ -0,0 +1,80 @@
+from phonemizer.phonemize import phonemize
+import re, regex
+
+
+## from Tacotron (from SIWIS corpus)
+punkt_class = "“”„«»!,.:;–?\"'…\[\](){}—\-" ## NOTE: without the whitespace
+re_punkt_class = re.compile("([" + punkt_class + "])")
+
+
+def espeak_phon(
+ text,
+ lang="ru",
+ accent=None,
+ clean=True,
+ add_punkt=True,
+ mark_pausation=True,
+ pausation_symbol="%",
+ _spec_seq="@@@",
+ verbose=False,
+):
+ language = lang
+ if accent:
+ language += f"-{accent}"
+ if verbose:
+ print(f"Original input:\n{text}\n")
+
+ punkt = None
+ if add_punkt:
+ spec_phonseq = phonemize(_spec_seq, language=language, backend="espeak").strip()
+ assert len(spec_phonseq) > 0
+ assert not re.match("\s", spec_phonseq)
+ punkt = re_punkt_class.finditer(text)
+ # punkt = re_punkt_class.findall(text)
+ # punkt = regex.findall('([' + punkt_class + '])', text)
+ punkt = [x.group(0) for x in punkt]
+ if verbose:
+ print(f"Punctuation recognized:\n{punkt}\n")
+ if len(punkt) > 0:
+ punkt = iter(punkt)
+ else:
+ punkt = None
+ text = re.sub(re_punkt_class, _spec_seq + r"\1", text)
+ if verbose:
+ print(f"Input with marked punctuation:\n{text}\n")
+
+ phoneseq = phonemize(text, language=language, backend="espeak")
+ if mark_pausation:
+ phoneseq = re.sub("\n", pausation_symbol, phoneseq)
+ else:
+ phoneseq = re.sub("\n", " ", phoneseq)
+ if verbose:
+ print(f"Input with marked pausation:\n{phoneseq}\n")
+
+ if clean:
+ phoneseq = regex.sub(
+ "\(.+?\)", "", phoneseq
+ ) ## use regex for non-greedy matching
+ ## zz = liaison after z
+ phoneseq = re.sub("z+", "z", phoneseq)
+ if verbose:
+ print(f'Input after "cleaning":\n{phoneseq}\n')
+
+ out_phoneseq = ""
+ if add_punkt and (punkt is not None):
+ split_phoneseq = [x for x in re.split("(" + spec_phonseq + ")", phoneseq) if x]
+ for x in split_phoneseq:
+ if x == spec_phonseq:
+ out_phoneseq += next(punkt)
+ else:
+ out_phoneseq += x
+ else:
+ out_phoneseq = phoneseq
+ out_phoneseq = re.sub(r" %", r"%", out_phoneseq)
+ out_phoneseq = re.sub(r"([" + punkt_class + "]+)(%)", r"\2\1 ", out_phoneseq)
+ out_phoneseq = re.sub(r" %", r"%", out_phoneseq)
+ out_phoneseq = re.sub("\s+", " ", out_phoneseq)
+ if verbose:
+ print(f"Output:\n{out_phoneseq}")
+
+ return out_phoneseq
From be2dc5e7f7d81643c930a6332df9263ecd9d7181 Mon Sep 17 00:00:00 2001
From: ruk0sh <43880105+ruk0sh@users.noreply.github.com>
Date: Tue, 19 Apr 2022 20:39:20 +0300
Subject: [PATCH 07/17] Add script for creating phonematic lexicon from word
corpus with espeak backend
---
src/preprocessing/create_lexicon.py | 47 +++++++++++++++++++
.../{phonemizer.py => espeak_phon.py} | 3 ++
2 files changed, 50 insertions(+)
create mode 100644 src/preprocessing/create_lexicon.py
rename src/preprocessing/{phonemizer.py => espeak_phon.py} (96%)
diff --git a/src/preprocessing/create_lexicon.py b/src/preprocessing/create_lexicon.py
new file mode 100644
index 0000000..ceed1e4
--- /dev/null
+++ b/src/preprocessing/create_lexicon.py
@@ -0,0 +1,47 @@
+"""
+Util to create phonematic lexicon for given corpus.
+Espeak backend must be installed on the system.
+"""
+
+from pathlib import Path
+
+import click
+from phonemizer.phonemize import phonemize
+from phonemizer.separator import Separator
+from tqdm import tqdm
+
+ENCODING = "utf8"
+SEPARATOR = Separator(phone=" ", syllable="", word="")
+
+
+@click.command()
+@click.option(
+ "--input-path", type=Path, help="", required=True,
+)
+@click.option(
+ "--output-path",
+ type=Path,
+ default="russian-lexicon-espeak.txt",
+ help="Filepath to write lexicon.",
+ required=True,
+)
+def main(input_path: Path, output_path: Path) -> None:
+ with open(input_path, "r", encoding=ENCODING) as corpus_file:
+ with open(output_path, "w", encoding=ENCODING) as lexicon_file:
+ for word in tqdm(list(corpus_file.readlines())):
+ word = word.strip()
+ phones = phonemize(
+ word,
+ language="ru",
+ backend="espeak",
+ preserve_punctuation=True,
+ with_stress=True,
+ separator=SEPARATOR,
+ )
+ lexicon_line = f"{word}\t{phones}\n"
+ lexicon_file.write(lexicon_line)
+ print(f"Lexicon file saved at:\n{output_path}")
+
+
+if __name__ == "__main__":
+ main()
diff --git a/src/preprocessing/phonemizer.py b/src/preprocessing/espeak_phon.py
similarity index 96%
rename from src/preprocessing/phonemizer.py
rename to src/preprocessing/espeak_phon.py
index 445b9e5..82a9d1a 100644
--- a/src/preprocessing/phonemizer.py
+++ b/src/preprocessing/espeak_phon.py
@@ -1,4 +1,5 @@
from phonemizer.phonemize import phonemize
+from phonemizer.separator import Separator
import re, regex
@@ -24,6 +25,8 @@ def espeak_phon(
if verbose:
print(f"Original input:\n{text}\n")
+ separator = Separator(phone=" ", syllable="", word="\n")
+
punkt = None
if add_punkt:
spec_phonseq = phonemize(_spec_seq, language=language, backend="espeak").strip()
From e34a13afbb85629b03734bc7442f186b13cb5f22 Mon Sep 17 00:00:00 2001
From: ruk0sh <43880105+ruk0sh@users.noreply.github.com>
Date: Tue, 19 Apr 2022 22:51:07 +0300
Subject: [PATCH 08/17] Radically improve performance
---
src/preprocessing/create_lexicon.py | 59 ++++++++++++++++++++---------
1 file changed, 42 insertions(+), 17 deletions(-)
diff --git a/src/preprocessing/create_lexicon.py b/src/preprocessing/create_lexicon.py
index ceed1e4..02c1be6 100644
--- a/src/preprocessing/create_lexicon.py
+++ b/src/preprocessing/create_lexicon.py
@@ -8,38 +8,63 @@
import click
from phonemizer.phonemize import phonemize
from phonemizer.separator import Separator
-from tqdm import tqdm
ENCODING = "utf8"
-SEPARATOR = Separator(phone=" ", syllable="", word="")
+SEPARATOR = Separator(phone=" ", syllable="", word="\n")
+CORPUS_STRING_WORD_SEPARATOR = " "
@click.command()
@click.option(
- "--input-path", type=Path, help="", required=True,
+ "-i",
+ "--input-path",
+ type=Path,
+ help="Path to corpus file (1 word per line).",
+ required=True,
)
@click.option(
+ "-o",
"--output-path",
type=Path,
default="russian-lexicon-espeak.txt",
help="Filepath to write lexicon.",
- required=True,
+ required=False,
)
def main(input_path: Path, output_path: Path) -> None:
+
+ print("Reading corpus...", end=" ")
with open(input_path, "r", encoding=ENCODING) as corpus_file:
- with open(output_path, "w", encoding=ENCODING) as lexicon_file:
- for word in tqdm(list(corpus_file.readlines())):
- word = word.strip()
- phones = phonemize(
- word,
- language="ru",
- backend="espeak",
- preserve_punctuation=True,
- with_stress=True,
- separator=SEPARATOR,
- )
- lexicon_line = f"{word}\t{phones}\n"
- lexicon_file.write(lexicon_line)
+ corpus_string = CORPUS_STRING_WORD_SEPARATOR.join(
+ corpus_file.read().splitlines()
+ )
+ print("Done.")
+
+ print("Getting phonemizations...", end=" ")
+ phones_string = phonemize(
+ corpus_string,
+ language="ru",
+ backend="espeak",
+ preserve_punctuation=True,
+ with_stress=True,
+ strip=True,
+ separator=SEPARATOR,
+ )
+ print("Done.")
+
+ print("Creating lexicon...", end=" ")
+ corpus = corpus_string.split(CORPUS_STRING_WORD_SEPARATOR)
+ phones = phones_string.split(SEPARATOR.word)
+ assert len(corpus) == len(phones), (
+ f"# of words should match # of phones after phomenization"
+ f"but you have {len(corpus)} words and {len(phones)} phones"
+ )
+ lexicon = [f"{word}\t{phon}" for word, phon in zip(corpus, phones)]
+ lexicon_str = "\n".join(lexicon)
+ print("Done.")
+
+ output_path.parent.mkdir(exist_ok=True, parents=True)
+ with open(output_path, "w", encoding=ENCODING) as lexicon_file:
+ lexicon_file.write(lexicon_str)
print(f"Lexicon file saved at:\n{output_path}")
From 8263bb23fef714d4363954c564b38617f61162df Mon Sep 17 00:00:00 2001
From: ruk0sh <43880105+ruk0sh@users.noreply.github.com>
Date: Tue, 19 Apr 2022 23:16:08 +0300
Subject: [PATCH 09/17] Parametrize language
---
src/preprocessing/create_lexicon.py | 24 ++++++++++++++++++------
1 file changed, 18 insertions(+), 6 deletions(-)
diff --git a/src/preprocessing/create_lexicon.py b/src/preprocessing/create_lexicon.py
index 02c1be6..12fc223 100644
--- a/src/preprocessing/create_lexicon.py
+++ b/src/preprocessing/create_lexicon.py
@@ -30,7 +30,15 @@
help="Filepath to write lexicon.",
required=False,
)
-def main(input_path: Path, output_path: Path) -> None:
+@click.option(
+ "-l",
+ "--language",
+ type=str,
+ default="en-us",
+ help="Language in espeak format. See for more info: https://github.com/espeak-ng/espeak-ng/blob/master/docs/languages.md",
+ required=False,
+)
+def main(input_path: Path, output_path: Path, language: str) -> None:
print("Reading corpus...", end=" ")
with open(input_path, "r", encoding=ENCODING) as corpus_file:
@@ -38,11 +46,12 @@ def main(input_path: Path, output_path: Path) -> None:
corpus_file.read().splitlines()
)
print("Done.")
+ print(corpus_string[:80], end="\n\n")
print("Getting phonemizations...", end=" ")
phones_string = phonemize(
corpus_string,
- language="ru",
+ language=language,
backend="espeak",
preserve_punctuation=True,
with_stress=True,
@@ -50,22 +59,25 @@ def main(input_path: Path, output_path: Path) -> None:
separator=SEPARATOR,
)
print("Done.")
+ print(phones_string[:80], end="\n\n")
print("Creating lexicon...", end=" ")
corpus = corpus_string.split(CORPUS_STRING_WORD_SEPARATOR)
phones = phones_string.split(SEPARATOR.word)
assert len(corpus) == len(phones), (
- f"# of words should match # of phones after phomenization"
- f"but you have {len(corpus)} words and {len(phones)} phones"
+ f"# of words should match # of phones after phomenization\n"
+ f"but you have {len(corpus)} words and {len(phones)} phones\n"
+ f"First 5 words: {corpus[:5]}\n"
+ f"First 5 phones: {phones[:5]}\n"
)
lexicon = [f"{word}\t{phon}" for word, phon in zip(corpus, phones)]
lexicon_str = "\n".join(lexicon)
- print("Done.")
+ print("Done.", end="\n\n")
output_path.parent.mkdir(exist_ok=True, parents=True)
with open(output_path, "w", encoding=ENCODING) as lexicon_file:
lexicon_file.write(lexicon_str)
- print(f"Lexicon file saved at:\n{output_path}")
+ print(f"Lexicon file saved at:\n{output_path}", end="\n\n")
if __name__ == "__main__":
From aee0ea97c67f873aded65ea8be2f37da019db6e0 Mon Sep 17 00:00:00 2001
From: ruk0sh <43880105+ruk0sh@users.noreply.github.com>
Date: Wed, 20 Apr 2022 13:35:01 +0300
Subject: [PATCH 10/17] Add lexicon creation with espeak phonemizer step
---
russian_preprocessing/steps.sh | 9 +++++++--
1 file changed, 7 insertions(+), 2 deletions(-)
diff --git a/russian_preprocessing/steps.sh b/russian_preprocessing/steps.sh
index f10bf8c..c7a0595 100644
--- a/russian_preprocessing/steps.sh
+++ b/russian_preprocessing/steps.sh
@@ -53,13 +53,18 @@ export RUS_ESPEAK_LEXICON=/media/diskB/ruslan_a/models/mfa/rus-espeak-mfa/rus-mf
# 16069/16069 [00:00<00:00, 21508.15it/s]
echo -e "\n6.1) Creating word list from dataset"
-python src/preprocessing/create_corpus.py --input-dir $EMO_OUTPUT_DIR/text/raw --output-path $EMO_OUTPUT_DIR/meta/words.txt
+python src/preprocessing/create_corpus.py --input-dir $EMO_OUTPUT_DIR/mfa_inputs --output-path $EMO_OUTPUT_DIR/meta/words.txt
+python src/preprocessing/create_corpus.py --input-dir $NEUTRAL_OUTPUT_DIR/mfa_inputs --output-path $NEUTRAL_OUTPUT_DIR/meta/words.txt
# 25849/25983 [01:19<00:00, 326.82it/s]
-echo -e "\n6.2) Creating G2P lexicon from word list"
+echo -e "\n6.2.a) Creating G2P lexicon from word list with G2P"
mfa g2p -t mfa_tmp -j 32 --clean --overwrite models/g2p/russian_g2p.zip $EMO_OUTPUT_DIR/meta/words.txt models/mfa/russian_lexicon.txt
rm -rf mfa_tmp
+echo -e "\n6.2.b) Creating G2P lexicon from word list with espeak"
+python src/preprocessing/create_lexicon.py -i $EMO_OUTPUT_DIR/meta/words.txt -o $EMO_OUTPUT_DIR/meta/lexicon.txt -l ru
+python src/preprocessing/create_lexicon.py -i $NEUTRAL_OUTPUT_DIR/meta/words.txt -o $NEUTRAL_OUTPUT_DIR/meta/lexicon.txt -l ru
+
# 42it [00:10, 4.01it/s]
# 58it [00:12, 4.54it/s] mix
echo -e "\n7) MFA Preprocessing"
From de36f30b5185f08391e791b35417501a9234f8e0 Mon Sep 17 00:00:00 2001
From: ruk0sh <43880105+ruk0sh@users.noreply.github.com>
Date: Wed, 20 Apr 2022 15:42:43 +0300
Subject: [PATCH 11/17] fix g2p download links
---
models/download_g2p.sh | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/models/download_g2p.sh b/models/download_g2p.sh
index 2e1faf4..3124bf1 100644
--- a/models/download_g2p.sh
+++ b/models/download_g2p.sh
@@ -1,6 +1,6 @@
-G2P_ENGLISH_URL="https://github.com/MontrealCorpusTools/mfa-models/blob/main/g2p/english_g2p.zip"
+G2P_ENGLISH_URL="https://github.com/MontrealCorpusTools/mfa-models/raw/main/g2p/english_g2p.zip"
G2P_ENGLISH_PATH="models/en/g2p/english_g2p.zip"
-G2P_RUSSIAN_URL="https://github.com/MontrealCorpusTools/mfa-models/blob/main/g2p/russian_g2p.zip"
+G2P_RUSSIAN_URL="https://github.com/MontrealCorpusTools/mfa-models/raw/main/g2p/russian_g2p.zip"
G2P_RUSSIAN_PATH="models/ru/g2p/russian_g2p.zip"
echo Downloading English G2P model...
From c766102c9eafe6893f774916dfe718a4fca06507 Mon Sep 17 00:00:00 2001
From: ruk0sh <43880105+ruk0sh@users.noreply.github.com>
Date: Wed, 4 May 2022 19:48:39 +0300
Subject: [PATCH 12/17] Add notebook for HTML-with-wavs generation.
---
generate_listener_html.ipynb | 1596 ++++++++++++++++++++++++++++++++++
1 file changed, 1596 insertions(+)
create mode 100644 generate_listener_html.ipynb
diff --git a/generate_listener_html.ipynb b/generate_listener_html.ipynb
new file mode 100644
index 0000000..36192f4
--- /dev/null
+++ b/generate_listener_html.ipynb
@@ -0,0 +1,1596 @@
+{
+ "cells": [
+ {
+ "cell_type": "code",
+ "execution_count": 1,
+ "id": "60f2a2f5-75b6-4f29-bfa3-687283fc3edd",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "import csv\n",
+ "from itertools import combinations\n",
+ "from pathlib import Path\n",
+ "\n",
+ "import numpy as np\n",
+ "import pandas as pd\n",
+ "from IPython.display import HTML, Audio, display\n",
+ "\n",
+ "\n",
+ "pd.set_option('display.max_colwidth', None)\n",
+ "pd.set_option('display.colheader_justify', 'center')"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 2,
+ "id": "f0d165ce-a7bb-4e83-b37e-dc0692ed1089",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "texts = [\n",
+ " 'We got few vegetables and fruits , and became fish eaters .',\n",
+ " 'Fifty yards ahead of her were the first of the rocks .',\n",
+ " 'It seemed the ordained order of things that dogs should work .',\n",
+ " 'The journey was continued at dawn .',\n",
+ " 'Was it the rendezvous of those who were striving to work his ruin .',\n",
+ " 'A dead man is of no use on a plantation .',\n",
+ " 'The Claudine was leaving next morning for Honolulu .',\n",
+ " 'Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting.',\n",
+ " 'Different telescope designs perform differently, and have different strengths and weaknesses.',\n",
+ " 'Humans also judge distance by using the relative sizes of objects.',\n",
+ " 'If this is true then those who tend to think creatively, really are somehow different.',\n",
+ " 'But really in the grand scheme of things, this information is insignificant.',\n",
+ " 'He had a private jet with three king-size beds, expensive rugs, porcelain vases and a dining area.',\n",
+ " 'When I reached Atlanta my steadily increasing disappointment was not lessened. I found it a big, dull, red town.',\n",
+ " 'She woke Meg with a \"Merry Christmas\", and bade her see what was under her pillow. A green–covered book appeared, with the same picture inside, and a few words written by their mother, which made their one present very precious in their eyes.',\n",
+ " \"Does Jane know about your new job? No, and don't you dare tell her! She will be furious!\",\n",
+ " 'Does she like ice cream or sweets? She likes any kind of ice cream. Chocolate, vanilla, strawberry, banana, the one with caramel, coconut, any you can think of!',\n",
+ " 'Earthquakes damage all structures, including bridges.',\n",
+ " 'Luckily, this kind of collapse is relatively infrequent.',\n",
+ " 'Was it using language that caused their brains to develop?',\n",
+ " 'If everyone followed a similar plan, the results would be impressive.',\n",
+ " 'Next, the hero overcomes obstacles on the way to facing their greatest challenge.',\n",
+ " 'For more than two hundred years the pessimists have been winning the public debate.',\n",
+ " \"It's wearing me out trying to juggle work with looking after my children and my family.\",\n",
+ " 'Young people want to feel supported and appreciated by their company and their superiors.',\n",
+ " 'We start to see the evidence of early human civilisation, through cave paintings for example.',\n",
+ " 'In this culture a so-called \"smile of respect\" is seen as insincere and often regarded with suspicion.',\n",
+ " 'We can express complex thoughts, convey subtle emotions and communicate about some abstract concepts.',\n",
+ " 'The activists send a clear message to companies that people are no longer willing to accept the environmental and human cost of overconsumption.',\n",
+ " 'All this is thanks to his childhood in the mountains and to genetics, but it is his mental strength that sets him apart.'\n",
+ "]"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 3,
+ "id": "f0871808-7396-4a82-b529-404babb6be8d",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "huawei_phones = [\n",
+ " ' W IY1 G AA1 T F Y UW1 V EH1 JH T AH0 B AH0 L Z AH0 N D F R UW1 T S AH0 N D B IH0 K EY1 M F IH1 SH IY1 T ER0 Z ',\n",
+ " ' F IH1 F T IY0 Y AA1 R D Z AH0 HH EH1 D AH1 V HH ER1 W ER1 DH AH0 F ER1 S T AH1 V DH AH0 R AA1 K S ',\n",
+ " ' IH1 T S IY1 M D DH IY0 AO0 R D EY1 N D AO1 R D ER0 AH1 V TH IH1 NG Z DH AE1 T D AA1 G Z SH UH1 D W ER1 K ',\n",
+ " ' DH AH0 JH ER1 N IY0 W AA1 Z K AH0 N T IH1 N Y UW0 D AE1 T D AO1 N ',\n",
+ " ' W AA1 Z IH1 T DH AH0 R AA1 N D IH0 V UW2 AH1 V DH OW1 Z HH UW1 W ER1 S T R AY1 V IH0 NG T UW1 W ER1 K HH IH1 Z R UW1 AH0 N ',\n",
+ " ' AH0 D EH1 D M AE1 N IH1 Z AH1 V N OW1 Y UW1 S AA1 N AH0 P L AE2 N T EY1 SH AH0 N ',\n",
+ " ' DH AH0 K L AO0 D IY1 N W AA1 Z L IY1 V IH0 NG N EH1 K S T M AO1 R N IH0 NG F AO1 R HH AA2 N AH0 L UW1 L UW0 ',\n",
+ " ' P R AA1 S IH0 K Y UW2 T ER0 Z HH AE1 V OW1 P AH0 N D AH0 M AE1 S IH0 V IH2 N V EH2 S T AH0 G EY1 SH AH0 N IH1 N T UW0 AE2 L AH0 G EY1 SH AH0 N Z AH1 V F IH1 K S IH0 NG G EY1 M Z AH0 N D IH2 L IY1 G AH0 L B EH1 T IH0 NG ',\n",
+ " ' D IH1 F ER0 AH0 N T T EH1 L AH0 S K OW2 P D IH0 Z AY1 N Z P ER0 F AO1 R M D IH1 F R AH0 N T L IY0 AH0 N D HH AE1 V D IH1 F ER0 AH0 N T S T R EH1 NG K TH S AH0 N D W IY1 K N AH0 S AH0 Z ',\n",
+ " ' HH Y UW1 M AH0 N Z AO1 L S OW0 JH AH1 JH D IH1 S T AH0 N S B AY1 Y UW1 Z IH0 NG DH AH0 R EH1 L AH0 T IH0 V S AY1 Z AH0 Z AH1 V AA1 B JH EH0 K T S ',\n",
+ " ' IH1 F DH IH1 S IH1 Z T R UW1 DH EH1 N DH OW1 Z HH UW1 T EH1 N D T UW1 TH IH1 NG K K R IY0 EY1 T IH0 V L IY0 R IH1 L IY0 AA1 R S AH1 M HH AW2 D IH1 F ER0 AH0 N T ',\n",
+ " ' B AH1 T R IH1 L IY0 IH0 N DH AH0 G R AE1 N D S K IY1 M AH1 V TH IH1 NG Z DH IH1 S IH2 N F ER0 M EY1 SH AH0 N IH1 Z IH2 N S IH0 G N Y IH1 F IH0 K AH0 N T ',\n",
+ " ' HH IY1 HH AE1 D AH0 P R AY1 V AH0 T JH EH1 T W IH1 DH TH R IY1 K IH1 NG S AY1 Z B EH1 D Z IH0 K S P EH1 N S IH0 V R AH1 G Z P AO1 R S AH0 L AH0 N V EY1 S AH0 Z AH0 N D AH0 D AY1 N IH0 NG EH1 R IY0 AH0 ',\n",
+ " ' W EH1 N AY1 R IY1 CH T AE0 T L AE1 N T AH0 M AY1 S T EH1 D AH0 L IY0 IH2 N K R IY1 S IH0 NG D IH2 S AH0 P OY1 N T M AH0 N T W AA1 Z N AA1 T L EH1 S AH0 N D AY1 F AW1 N D IH1 T AH0 B IH1 G D AH1 L R EH1 D T AW1 N ',\n",
+ " ' SH IY1 W OW1 K M EH1 G W IH1 DH AH0 M EH1 R IY0 K R IH1 S M AH0 S AH0 N D B EY1 D HH ER1 S IY1 W AH1 T W AA1 Z AH1 N D ER0 HH ER1 P IH1 L OW0 AH0 G R IY1 N K AH1 V ER0 D B UH1 K AH0 P IH1 R D W IH1 DH DH AH0 S EY1 M P IH1 K CH ER0 IH2 N S AY1 D AH0 N D AH0 F Y UW1 W ER1 D Z R IH1 T AH0 N B AY1 DH EH1 R M AH1 DH ER0 W IH1 CH M EY1 D DH EH1 R W AH1 N P R EH1 Z AH0 N T V EH1 R IY0 P R EH1 SH AH0 S IH0 N DH EH1 R AY1 Z ',\n",
+ " ' D AH1 Z JH EY1 N N OW1 AH0 B AW1 T Y AO1 R N UW1 JH AA1 B N OW1 AH0 N D D OW1 N T Y UW1 D EH1 R T EH1 L HH ER1 SH IY1 W IH1 L B IY1 F Y UH1 R IY0 AH0 S ',\n",
+ " ' D AH1 Z SH IY1 L AY1 K AY1 S K R IY1 M AO1 R S W IY1 T S SH IY1 L AY1 K S EH1 N IY0 K AY1 N D AH1 V AY1 S K R IY1 M CH AO1 K L AH0 T V AH0 N IH1 L AH0 S T R AO1 B EH2 R IY0 B AH0 N AE1 N AH0 DH AH0 W AH1 N W IH1 DH K EH1 R AH0 M AH0 L K OW1 K AH0 N AH2 T EH1 N IY0 Y UW1 K AE1 N TH IH1 NG K AH1 V ',\n",
+ " ' ER1 TH K W EY2 K S D AE1 M AH0 JH AO1 L S T R AH1 K CH ER0 Z IH2 N K L UW1 D IH0 NG B R IH1 JH AH0 Z ',\n",
+ " ' L AH1 K AH0 L IY0 DH IH1 S K AY1 N D AH1 V K AH0 L AE1 P S IH1 Z R EH1 L AH0 T IH0 V L IY0 IH2 N F R IY1 K W AH0 N T ',\n",
+ " ' W AA1 Z IH1 T Y UW1 Z IH0 NG L AE1 NG G W AH0 JH DH AE1 T K AA1 Z D DH EH1 R B R EY1 N Z T UW1 D IH0 V EH1 L AH0 P ',\n",
+ " ' IH1 F EH1 V R IY0 W AH2 N F AA1 L OW0 D AH0 S IH1 M AH0 L ER0 P L AE1 N DH AH0 R IH0 Z AH1 L T S W UH1 D B IY1 IH2 M P R EH1 S IH0 V ',\n",
+ " ' N EH1 K S T DH AH0 HH IH1 R OW0 OW1 V ER0 K AH2 M Z AA1 B S T AH0 K AH0 L Z AA1 N DH AH0 W EY1 T UW1 F EY1 S IH0 NG DH EH1 R G R EY1 T AH0 S T CH AE1 L AH0 N JH ',\n",
+ " ' F AO1 R M AO1 R DH AE1 N T UW1 HH AH1 N D R AH0 D Y IH1 R Z DH AH0 P EH1 S AH0 M IH0 S T S HH AE1 V B IH1 N W IH1 N IH0 NG DH AH0 P AH1 B L IH0 K D AH0 B EY1 T ',\n",
+ " ' IH1 T S W EH1 R IH0 NG M IY1 AW1 T T R AY1 IH0 NG T UW1 JH AH1 G AH0 L W ER1 K W IH1 DH L UH1 K IH0 NG AE1 F T ER0 M AY1 CH IH1 L D R AH0 N AH0 N D M AY1 F AE1 M AH0 L IY0 ',\n",
+ " ' Y AH1 NG P IY1 P AH0 L W AA1 N T T UW1 F IY1 L S AH0 P AO1 R T IH0 D AH0 N D AH0 P R IY1 SH IY0 EY2 T IH0 D B AY1 DH EH1 R K AH1 M P AH0 N IY2 AH0 N D DH EH1 R S UW0 P IH1 R IY0 ER0 Z ',\n",
+ " ' W IY1 S T AA1 R T T UW1 S IY1 DH IY0 EH1 V AH0 D AH0 N S AH1 V ER1 L IY0 HH Y UW1 M AH0 N S IH1 V AH0 L IH0 S EY1 SH AH0 N TH R UW1 K EY1 V P EY1 N T IH0 NG Z F AO1 R IH0 G Z AE1 M P AH0 L ',\n",
+ " ' IH0 N DH IH1 S K AH1 L CH ER0 AH0 S OW1 K AO1 L D S M AY1 L AH1 V R IH0 S P EH1 K T IH1 Z S IY1 N AE1 Z IH2 N S IH0 N S IH1 R AH0 N D AO1 F AH0 N R IH0 G AA1 R D IH0 D W IH1 DH S AH0 S P IH1 SH AH0 N ',\n",
+ " ' W IY1 K AE1 N IH0 K S P R EH1 S K AA1 M P L EH0 K S TH AO1 T S K AH0 N V EY1 S AH1 T AH0 L IH0 M OW1 SH AH0 N Z AH0 N D K AH0 M Y UW1 N AH0 K EY2 T AH0 B AW1 T S AH1 M AE1 B S T R AE0 K T K AA1 N S EH0 P T S ',\n",
+ " ' DH IY0 AE1 K T AH0 V AH0 S T S S EH1 N D AH0 K L IH1 R M EH1 S AH0 JH T UW1 K AH1 M P AH0 N IY2 Z DH AE1 T P IY1 P AH0 L AA1 R N OW1 L AO1 NG G ER0 W IH1 L IH0 NG T UW1 AE0 K S EH1 P T DH IY0 IH0 N V AY2 R AH0 N M EH1 N T AH0 L AH0 N D HH Y UW1 M AH0 N K AA1 S T AH1 V OW1 V ER0 K AH0 N S AH2 M P SH AH0 N ',\n",
+ " ' AO1 L DH IH1 S IH1 Z TH AE1 NG K S T UW1 HH IH1 Z CH AY1 L D HH UH2 D IH0 N DH AH0 M AW1 N T AH0 N Z AH0 N D T UW1 JH AH0 N EH1 T IH0 K S B AH1 T IH1 T IH1 Z HH IH1 Z M EH1 N T AH0 L S T R EH1 NG K TH DH AE1 T S EH1 T S HH IH1 M AH0 P AA1 R T '\n",
+ "]"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 4,
+ "id": "d7942da8-a539-4a50-98ef-d93cbdcec403",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "## NOTE: all model keys need to be composed of a single letter followed by a number\n",
+ "model2folder = {\n",
+ " \"m1\": 'generated_hifi/esd_tune/g_3164999',\n",
+ " \"m2\": 'generated_hifi/esd_tune_reversal/g_3164999',\n",
+ " \"m3\": 'generated_hifi/esd_tune_advloss0/g_3164999',\n",
+ "}"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 5,
+ "id": "d4d44f9e-525a-4e3d-bb5c-679b64bd7861",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "total_K_models = len(model2folder)"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 6,
+ "id": "51e4eae6-ea64-4d69-947c-2a0c7b15678b",
+ "metadata": {},
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "m1: 1350 wav files total\n",
+ "m2: 1350 wav files total\n",
+ "m3: 1350 wav files total\n"
+ ]
+ }
+ ],
+ "source": [
+ "wav_files = {}\n",
+ "total_wavs_per_model = 0\n",
+ "for ip in model2folder.keys():\n",
+ " wav_files[ip] = list(Path(model2folder[ip]).rglob(\"*.wav\"))\n",
+ " total_wavs_per_model = len(wav_files[ip])\n",
+ " print(f\"{ip}: {total_wavs_per_model} wav files total\")"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 7,
+ "id": "e2031149-e1e8-421f-983c-83833c1b67ad",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "x_range = np.array(list(range(total_wavs_per_model)))\n",
+ "x_indices = x_range * 2\n",
+ "y_indices = x_range * 2 + 1"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 8,
+ "id": "f60b3404-269a-407e-bedd-898b754c5651",
+ "metadata": {},
+ "outputs": [
+ {
+ "data": {
+ "text/html": [
+ "\n",
+ "\n",
+ "
\n",
+ " \n",
+ " \n",
+ " | \n",
+ " m1 | \n",
+ " m2 | \n",
+ " m3 | \n",
+ " m1_wavbasename | \n",
+ " m1_accent_speaker | \n",
+ " m1_text | \n",
+ " m2_wavbasename | \n",
+ " m2_accent_speaker | \n",
+ " m2_text | \n",
+ " m3_wavbasename | \n",
+ " m3_accent_speaker | \n",
+ " m3_text | \n",
+ "
\n",
+ " \n",
+ " \n",
+ " \n",
+ " | 0 | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ "
\n",
+ " \n",
+ " | 1 | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ "
\n",
+ " \n",
+ " | 2 | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ "
\n",
+ " \n",
+ " | 3 | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ "
\n",
+ " \n",
+ " | 4 | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ "
\n",
+ " \n",
+ " | ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ "
\n",
+ " \n",
+ " | 2695 | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ "
\n",
+ " \n",
+ " | 2696 | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ "
\n",
+ " \n",
+ " | 2697 | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ "
\n",
+ " \n",
+ " | 2698 | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ "
\n",
+ " \n",
+ " | 2699 | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
2700 rows × 12 columns
\n",
+ "
\n",
+ "\n",
+ "
\n",
+ " \n",
+ " \n",
+ " | \n",
+ " m1 | \n",
+ " m2 | \n",
+ " m3 | \n",
+ " m1_wavbasename | \n",
+ " m1_accent_speaker | \n",
+ " m1_text | \n",
+ " m2_wavbasename | \n",
+ " m2_accent_speaker | \n",
+ " m2_text | \n",
+ " m3_wavbasename | \n",
+ " m3_accent_speaker | \n",
+ " m3_text | \n",
+ "
\n",
+ " \n",
+ " \n",
+ " \n",
+ " | 0 | \n",
+ " id_1 | \n",
+ " id_2 | \n",
+ " id_3 | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ "
\n",
+ " \n",
+ " | 1 | \n",
+ " generated_hifi/esd_tune/g_3164999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_3164999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_3164999/0011/angry/1.wav | \n",
+ " 1.wav | \n",
+ " 0011 | \n",
+ " We got few vegetables and fruits , and became fish eaters . | \n",
+ " 1.wav | \n",
+ " 0011 | \n",
+ " We got few vegetables and fruits , and became fish eaters . | \n",
+ " 1.wav | \n",
+ " 0011 | \n",
+ " We got few vegetables and fruits , and became fish eaters . | \n",
+ "
\n",
+ " \n",
+ " | 2 | \n",
+ " id_4 | \n",
+ " id_5 | \n",
+ " id_6 | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ "
\n",
+ " \n",
+ " | 3 | \n",
+ " generated_hifi/esd_tune/g_3164999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_3164999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_3164999/0011/angry/10.wav | \n",
+ " 10.wav | \n",
+ " 0011 | \n",
+ " Humans also judge distance by using the relative sizes of objects. | \n",
+ " 10.wav | \n",
+ " 0011 | \n",
+ " Humans also judge distance by using the relative sizes of objects. | \n",
+ " 10.wav | \n",
+ " 0011 | \n",
+ " Humans also judge distance by using the relative sizes of objects. | \n",
+ "
\n",
+ " \n",
+ " | 4 | \n",
+ " id_7 | \n",
+ " id_8 | \n",
+ " id_9 | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ "
\n",
+ " \n",
+ " | ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ "
\n",
+ " \n",
+ " | 2695 | \n",
+ " generated_hifi/esd_tune/g_3164999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/7.wav | \n",
+ " 7.wav | \n",
+ " 0020 | \n",
+ " The Claudine was leaving next morning for Honolulu . | \n",
+ " 7.wav | \n",
+ " 0020 | \n",
+ " The Claudine was leaving next morning for Honolulu . | \n",
+ " 7.wav | \n",
+ " 0020 | \n",
+ " The Claudine was leaving next morning for Honolulu . | \n",
+ "
\n",
+ " \n",
+ " | 2696 | \n",
+ " id_4045 | \n",
+ " id_4046 | \n",
+ " id_4047 | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ "
\n",
+ " \n",
+ " | 2697 | \n",
+ " generated_hifi/esd_tune/g_3164999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/8.wav | \n",
+ " 8.wav | \n",
+ " 0020 | \n",
+ " Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. | \n",
+ " 8.wav | \n",
+ " 0020 | \n",
+ " Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. | \n",
+ " 8.wav | \n",
+ " 0020 | \n",
+ " Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. | \n",
+ "
\n",
+ " \n",
+ " | 2698 | \n",
+ " id_4048 | \n",
+ " id_4049 | \n",
+ " id_4050 | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ " None | \n",
+ "
\n",
+ " \n",
+ " | 2699 | \n",
+ " generated_hifi/esd_tune/g_3164999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/9.wav | \n",
+ " 9.wav | \n",
+ " 0020 | \n",
+ " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ " 9.wav | \n",
+ " 0020 | \n",
+ " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ " 9.wav | \n",
+ " 0020 | \n",
+ " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
2700 rows × 12 columns
\n",
+ "
\n",
+ "\n",
+ "
\n",
+ " \n",
+ " \n",
+ " | \n",
+ " m1_text | \n",
+ " m1_accent_speaker | \n",
+ " m1 | \n",
+ " m2 | \n",
+ " m3 | \n",
+ " m2_text | \n",
+ "
\n",
+ " \n",
+ " \n",
+ " \n",
+ " | 0 | \n",
+ " None | \n",
+ " None | \n",
+ " id_1 | \n",
+ " id_2 | \n",
+ " id_3 | \n",
+ " None | \n",
+ "
\n",
+ " \n",
+ " | 1 | \n",
+ " We got few vegetables and fruits , and became fish eaters . | \n",
+ " 0011 | \n",
+ " generated_hifi/esd_tune/g_3164999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_3164999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_3164999/0011/angry/1.wav | \n",
+ " We got few vegetables and fruits , and became fish eaters . | \n",
+ "
\n",
+ " \n",
+ " | 2 | \n",
+ " None | \n",
+ " None | \n",
+ " id_4 | \n",
+ " id_5 | \n",
+ " id_6 | \n",
+ " None | \n",
+ "
\n",
+ " \n",
+ " | 3 | \n",
+ " Humans also judge distance by using the relative sizes of objects. | \n",
+ " 0011 | \n",
+ " generated_hifi/esd_tune/g_3164999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_3164999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_3164999/0011/angry/10.wav | \n",
+ " Humans also judge distance by using the relative sizes of objects. | \n",
+ "
\n",
+ " \n",
+ " | 4 | \n",
+ " None | \n",
+ " None | \n",
+ " id_7 | \n",
+ " id_8 | \n",
+ " id_9 | \n",
+ " None | \n",
+ "
\n",
+ " \n",
+ " | ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ "
\n",
+ " \n",
+ " | 2695 | \n",
+ " The Claudine was leaving next morning for Honolulu . | \n",
+ " 0020 | \n",
+ " generated_hifi/esd_tune/g_3164999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/7.wav | \n",
+ " The Claudine was leaving next morning for Honolulu . | \n",
+ "
\n",
+ " \n",
+ " | 2696 | \n",
+ " None | \n",
+ " None | \n",
+ " id_4045 | \n",
+ " id_4046 | \n",
+ " id_4047 | \n",
+ " None | \n",
+ "
\n",
+ " \n",
+ " | 2697 | \n",
+ " Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. | \n",
+ " 0020 | \n",
+ " generated_hifi/esd_tune/g_3164999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/8.wav | \n",
+ " Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. | \n",
+ "
\n",
+ " \n",
+ " | 2698 | \n",
+ " None | \n",
+ " None | \n",
+ " id_4048 | \n",
+ " id_4049 | \n",
+ " id_4050 | \n",
+ " None | \n",
+ "
\n",
+ " \n",
+ " | 2699 | \n",
+ " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ " 0020 | \n",
+ " generated_hifi/esd_tune/g_3164999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/9.wav | \n",
+ " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
2700 rows × 6 columns
\n",
+ "
\n",
+ "\n",
+ "
\n",
+ " \n",
+ " \n",
+ " | \n",
+ " m1_text | \n",
+ " m1_accent_speaker | \n",
+ " m1 | \n",
+ " m2 | \n",
+ " m3 | \n",
+ " m2_text | \n",
+ "
\n",
+ " \n",
+ " \n",
+ " \n",
+ " | 0 | \n",
+ " - | \n",
+ " - | \n",
+ " id_1 | \n",
+ " id_2 | \n",
+ " id_3 | \n",
+ " - | \n",
+ "
\n",
+ " \n",
+ " | 1 | \n",
+ " We got few vegetables and fruits , and became fish eaters . | \n",
+ " 0011 | \n",
+ " generated_hifi/esd_tune_reversal/g_3164999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune/g_3164999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_3164999/0011/angry/1.wav | \n",
+ " We got few vegetables and fruits , and became fish eaters . | \n",
+ "
\n",
+ " \n",
+ " | 2 | \n",
+ " - | \n",
+ " - | \n",
+ " id_4 | \n",
+ " id_5 | \n",
+ " id_6 | \n",
+ " - | \n",
+ "
\n",
+ " \n",
+ " | 3 | \n",
+ " Humans also judge distance by using the relative sizes of objects. | \n",
+ " 0011 | \n",
+ " generated_hifi/esd_tune_reversal/g_3164999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_3164999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune/g_3164999/0011/angry/10.wav | \n",
+ " Humans also judge distance by using the relative sizes of objects. | \n",
+ "
\n",
+ " \n",
+ " | 4 | \n",
+ " - | \n",
+ " - | \n",
+ " id_7 | \n",
+ " id_8 | \n",
+ " id_9 | \n",
+ " - | \n",
+ "
\n",
+ " \n",
+ " | ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ "
\n",
+ " \n",
+ " | 2695 | \n",
+ " The Claudine was leaving next morning for Honolulu . | \n",
+ " 0020 | \n",
+ " generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune/g_3164999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/7.wav | \n",
+ " The Claudine was leaving next morning for Honolulu . | \n",
+ "
\n",
+ " \n",
+ " | 2696 | \n",
+ " - | \n",
+ " - | \n",
+ " id_4045 | \n",
+ " id_4046 | \n",
+ " id_4047 | \n",
+ " - | \n",
+ "
\n",
+ " \n",
+ " | 2697 | \n",
+ " Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. | \n",
+ " 0020 | \n",
+ " generated_hifi/esd_tune/g_3164999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/8.wav | \n",
+ " Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. | \n",
+ "
\n",
+ " \n",
+ " | 2698 | \n",
+ " - | \n",
+ " - | \n",
+ " id_4048 | \n",
+ " id_4049 | \n",
+ " id_4050 | \n",
+ " - | \n",
+ "
\n",
+ " \n",
+ " | 2699 | \n",
+ " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ " 0020 | \n",
+ " generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune/g_3164999/0020/surprise/9.wav | \n",
+ " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
2700 rows × 6 columns
\n",
+ "
{name} | \" for name in column_names])\n",
+ " table = [f\"{names}
\"]\n",
+ " for row in zip(*columns):\n",
+ " row = \"\\n\".join([f\"{elem} | \" for elem in row])\n",
+ " row = f'{row}
'\n",
+ " table.append(row)\n",
+ " \n",
+ " content = \"\\n\".join(table)\n",
+ " table_html = f''\n",
+ " \n",
+ " html = HTML(table_html)\n",
+ " # display(html)\n",
+ " return html"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 17,
+ "id": "11d84c76-841a-4215-a45f-268076e4ae21",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "def make_table_from_df(df, column_names):\n",
+ " columns, column_names = columns_from_filenames(df, column_names)\n",
+ " return make_table(columns, column_names)"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 18,
+ "id": "26b577fe-3957-4902-8758-e5501f5d81a0",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "column_names = df_permuted.columns\n",
+ "chunk_size = 100"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 19,
+ "id": "f37b7fb3-93eb-448c-a8a3-a23cb0d165ae",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "output_folder = Path('reports/html')\n",
+ "output_folder.mkdir(parents=True, exist_ok=True)"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 20,
+ "id": "24aef771-215a-4fe6-82cb-1ff375681189",
+ "metadata": {
+ "scrolled": true,
+ "tags": []
+ },
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "0 100\n",
+ "reports/html/nat_vctk_esd_tune_test_1.html\n",
+ "100 200\n",
+ "reports/html/nat_vctk_esd_tune_test_2.html\n",
+ "200 300\n",
+ "reports/html/nat_vctk_esd_tune_test_3.html\n",
+ "300 400\n",
+ "reports/html/nat_vctk_esd_tune_test_4.html\n",
+ "400 500\n",
+ "reports/html/nat_vctk_esd_tune_test_5.html\n",
+ "500 600\n",
+ "reports/html/nat_vctk_esd_tune_test_6.html\n",
+ "600 700\n",
+ "reports/html/nat_vctk_esd_tune_test_7.html\n",
+ "700 800\n",
+ "reports/html/nat_vctk_esd_tune_test_8.html\n",
+ "800 900\n",
+ "reports/html/nat_vctk_esd_tune_test_9.html\n",
+ "900 1000\n",
+ "reports/html/nat_vctk_esd_tune_test_10.html\n",
+ "1000 1100\n",
+ "reports/html/nat_vctk_esd_tune_test_11.html\n",
+ "1100 1200\n",
+ "reports/html/nat_vctk_esd_tune_test_12.html\n",
+ "1200 1300\n",
+ "reports/html/nat_vctk_esd_tune_test_13.html\n",
+ "1300 1400\n",
+ "reports/html/nat_vctk_esd_tune_test_14.html\n",
+ "1400 1500\n",
+ "reports/html/nat_vctk_esd_tune_test_15.html\n",
+ "1500 1600\n",
+ "reports/html/nat_vctk_esd_tune_test_16.html\n",
+ "1600 1700\n",
+ "reports/html/nat_vctk_esd_tune_test_17.html\n",
+ "1700 1800\n",
+ "reports/html/nat_vctk_esd_tune_test_18.html\n",
+ "1800 1900\n",
+ "reports/html/nat_vctk_esd_tune_test_19.html\n",
+ "1900 2000\n",
+ "reports/html/nat_vctk_esd_tune_test_20.html\n",
+ "2000 2100\n",
+ "reports/html/nat_vctk_esd_tune_test_21.html\n",
+ "2100 2200\n",
+ "reports/html/nat_vctk_esd_tune_test_22.html\n",
+ "2200 2300\n",
+ "reports/html/nat_vctk_esd_tune_test_23.html\n",
+ "2300 2400\n",
+ "reports/html/nat_vctk_esd_tune_test_24.html\n",
+ "2400 2500\n",
+ "reports/html/nat_vctk_esd_tune_test_25.html\n",
+ "2500 2600\n",
+ "reports/html/nat_vctk_esd_tune_test_26.html\n",
+ "2600 2700\n",
+ "reports/html/nat_vctk_esd_tune_test_27.html\n"
+ ]
+ }
+ ],
+ "source": [
+ "def create_htmls(df, column_names, chunk_size):\n",
+ " import os\n",
+ " chunk_number = 0\n",
+ " start = 0\n",
+ " experiment_prefix = \"nat_vctk_esd_tune_test_\"\n",
+ " while (start < len(df)):\n",
+ " end = start + chunk_size\n",
+ " end = min(end, len(df))\n",
+ " print(start, end)\n",
+ " chunk_number += 1\n",
+ " df_chunk = df.iloc[start:end,:]\n",
+ " final_html = make_table_from_df(df_chunk, column_names)\n",
+ " f_html = output_folder.joinpath(experiment_prefix + str(chunk_number) + '.html')\n",
+ " print(f_html)\n",
+ " with open(f_html, 'w') as f:\n",
+ " f.write(final_html.data)\n",
+ " start = end\n",
+ "\n",
+ " \n",
+ "create_htmls(df_permuted, column_names, chunk_size)"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 21,
+ "id": "49a60285-31bf-4263-9162-c3d928aa9068",
+ "metadata": {},
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "reports/scores/scores.csv\n",
+ "reports/scores/scores.xlsx\n"
+ ]
+ }
+ ],
+ "source": [
+ "def create_scores_file(df, output_dir):\n",
+ " \n",
+ " output_dir = Path(output_dir)\n",
+ " output_dir.mkdir(exist_ok=True, parents=True)\n",
+ " \n",
+ " for k in model2folder.keys():\n",
+ " df[k].iloc[y_indices] = ''\n",
+ " \n",
+ " csv_path = output_dir / \"scores.csv\"\n",
+ " df.to_csv(\n",
+ " csv_path,\n",
+ " sep = \"\\t\",\n",
+ " header = True,\n",
+ " index = False,\n",
+ " quoting = csv.QUOTE_NONE,\n",
+ " )\n",
+ " \n",
+ " excel_path = output_dir / \"scores.xlsx\" \n",
+ " df.to_excel(excel_path)\n",
+ " \n",
+ " print(csv_path)\n",
+ " print(excel_path)\n",
+ "\n",
+ "\n",
+ "create_scores_file(df_permuted, \"reports/scores\")"
+ ]
+ }
+ ],
+ "metadata": {
+ "kernelspec": {
+ "display_name": "Python 3 (ipykernel)",
+ "language": "python",
+ "name": "python3"
+ },
+ "language_info": {
+ "codemirror_mode": {
+ "name": "ipython",
+ "version": 3
+ },
+ "file_extension": ".py",
+ "mimetype": "text/x-python",
+ "name": "python",
+ "nbconvert_exporter": "python",
+ "pygments_lexer": "ipython3",
+ "version": "3.8.12"
+ }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}
From f1996df97f4dd40fa46f84cd207e9de1a8e423ef Mon Sep 17 00:00:00 2001
From: ruk0sh <43880105+ruk0sh@users.noreply.github.com>
Date: Thu, 5 May 2022 13:07:27 +0300
Subject: [PATCH 13/17] Add notebook for analyzing gathered MOS/pref scores.
---
analyze_scores.ipynb | 13171 +++++++++++++++++++++++++++++++++++++++++
1 file changed, 13171 insertions(+)
create mode 100644 analyze_scores.ipynb
diff --git a/analyze_scores.ipynb b/analyze_scores.ipynb
new file mode 100644
index 0000000..e62b14a
--- /dev/null
+++ b/analyze_scores.ipynb
@@ -0,0 +1,13171 @@
+{
+ "cells": [
+ {
+ "cell_type": "code",
+ "execution_count": 1,
+ "id": "9169e316-80c7-460e-86d1-bca78ce044d3",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "import csv\n",
+ "from pathlib import Path\n",
+ "\n",
+ "import numpy as np\n",
+ "import pandas as pd"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 2,
+ "id": "9573a7f9-058c-47af-9af1-50fb4d454abb",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "pd.set_option('display.max_colwidth', None)\n",
+ "pd.set_option('display.colheader_justify', 'center')"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 3,
+ "id": "a199810d-893f-494d-b560-a1a82612bd13",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "SAMPLE_AUDIO_PATH = \"generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/9.wav\"\n",
+ "EVAL_MAP_PATH = \"reports/eval_map.csv\"\n",
+ "SCORES_PATH = \"data/scores_results.xlsx\""
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 4,
+ "id": "ac502266-1c5e-4dce-add2-a57d107571ed",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "## NOTE: all model keys need to be composed of a single letter followed by a number\n",
+ "model2folder = {\n",
+ " \"m1\": 'generated_hifi/esd_tune/g_3164999',\n",
+ " \"m2\": 'generated_hifi/esd_tune_reversal/g_3164999',\n",
+ " \"m3\": 'generated_hifi/esd_tune_advloss0/g_3164999',\n",
+ "}"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 5,
+ "id": "b79125a1-4e4a-483a-a498-ea95ed0eb61d",
+ "metadata": {},
+ "outputs": [
+ {
+ "data": {
+ "text/plain": [
+ "{'generated_hifi/esd_tune/g_3164999': 'm1',\n",
+ " 'generated_hifi/esd_tune_reversal/g_3164999': 'm2',\n",
+ " 'generated_hifi/esd_tune_advloss0/g_3164999': 'm3'}"
+ ]
+ },
+ "execution_count": 5,
+ "metadata": {},
+ "output_type": "execute_result"
+ }
+ ],
+ "source": [
+ "folder2model = {v: k for k, v in model2folder.items()}\n",
+ "folder2model"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 6,
+ "id": "8a2711f2-14ea-48f8-b8a4-bd6e5d845a0d",
+ "metadata": {},
+ "outputs": [
+ {
+ "data": {
+ "text/html": [
+ "\n",
+ "\n",
+ "
\n",
+ " \n",
+ " \n",
+ " | \n",
+ " 0 | \n",
+ " 1 | \n",
+ " 2 | \n",
+ " 3 | \n",
+ " 4 | \n",
+ " 5 | \n",
+ "
\n",
+ " \n",
+ " \n",
+ " \n",
+ " | 0 | \n",
+ " - | \n",
+ " - | \n",
+ " id_1 | \n",
+ " id_2 | \n",
+ " id_3 | \n",
+ " - | \n",
+ "
\n",
+ " \n",
+ " | 1 | \n",
+ " We got few vegetables and fruits , and became fish eaters . | \n",
+ " 0011 | \n",
+ " generated_hifi/esd_tune_reversal/g_3164999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune/g_3164999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_3164999/0011/angry/1.wav | \n",
+ " We got few vegetables and fruits , and became fish eaters . | \n",
+ "
\n",
+ " \n",
+ " | 2 | \n",
+ " - | \n",
+ " - | \n",
+ " id_4 | \n",
+ " id_5 | \n",
+ " id_6 | \n",
+ " - | \n",
+ "
\n",
+ " \n",
+ " | 3 | \n",
+ " Humans also judge distance by using the relative sizes of objects. | \n",
+ " 0011 | \n",
+ " generated_hifi/esd_tune_reversal/g_3164999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_3164999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune/g_3164999/0011/angry/10.wav | \n",
+ " Humans also judge distance by using the relative sizes of objects. | \n",
+ "
\n",
+ " \n",
+ " | 4 | \n",
+ " - | \n",
+ " - | \n",
+ " id_7 | \n",
+ " id_8 | \n",
+ " id_9 | \n",
+ " - | \n",
+ "
\n",
+ " \n",
+ " | ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ "
\n",
+ " \n",
+ " | 2695 | \n",
+ " The Claudine was leaving next morning for Honolulu . | \n",
+ " 0020 | \n",
+ " generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune/g_3164999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/7.wav | \n",
+ " The Claudine was leaving next morning for Honolulu . | \n",
+ "
\n",
+ " \n",
+ " | 2696 | \n",
+ " - | \n",
+ " - | \n",
+ " id_4045 | \n",
+ " id_4046 | \n",
+ " id_4047 | \n",
+ " - | \n",
+ "
\n",
+ " \n",
+ " | 2697 | \n",
+ " Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. | \n",
+ " 0020 | \n",
+ " generated_hifi/esd_tune/g_3164999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/8.wav | \n",
+ " Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. | \n",
+ "
\n",
+ " \n",
+ " | 2698 | \n",
+ " - | \n",
+ " - | \n",
+ " id_4048 | \n",
+ " id_4049 | \n",
+ " id_4050 | \n",
+ " - | \n",
+ "
\n",
+ " \n",
+ " | 2699 | \n",
+ " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ " 0020 | \n",
+ " generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune/g_3164999/0020/surprise/9.wav | \n",
+ " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
2700 rows × 6 columns
\n",
+ "
\n",
+ "\n",
+ "
\n",
+ " \n",
+ " \n",
+ " | \n",
+ " text_left | \n",
+ " speaker | \n",
+ " m1 | \n",
+ " m2 | \n",
+ " m3 | \n",
+ " text_right | \n",
+ "
\n",
+ " \n",
+ " \n",
+ " \n",
+ " | 0 | \n",
+ " - | \n",
+ " - | \n",
+ " id_1 | \n",
+ " id_2 | \n",
+ " id_3 | \n",
+ " - | \n",
+ "
\n",
+ " \n",
+ " | 1 | \n",
+ " We got few vegetables and fruits , and became fish eaters . | \n",
+ " 0011 | \n",
+ " {'model': 'm2', 'sent_id': 1, 'speaker': '0011', 'accent': None} | \n",
+ " {'model': 'm1', 'sent_id': 1, 'speaker': '0011', 'accent': None} | \n",
+ " {'model': 'm3', 'sent_id': 1, 'speaker': '0011', 'accent': None} | \n",
+ " We got few vegetables and fruits , and became fish eaters . | \n",
+ "
\n",
+ " \n",
+ " | 2 | \n",
+ " - | \n",
+ " - | \n",
+ " id_4 | \n",
+ " id_5 | \n",
+ " id_6 | \n",
+ " - | \n",
+ "
\n",
+ " \n",
+ " | 3 | \n",
+ " Humans also judge distance by using the relative sizes of objects. | \n",
+ " 0011 | \n",
+ " {'model': 'm2', 'sent_id': 10, 'speaker': '0011', 'accent': None} | \n",
+ " {'model': 'm3', 'sent_id': 10, 'speaker': '0011', 'accent': None} | \n",
+ " {'model': 'm1', 'sent_id': 10, 'speaker': '0011', 'accent': None} | \n",
+ " Humans also judge distance by using the relative sizes of objects. | \n",
+ "
\n",
+ " \n",
+ " | 4 | \n",
+ " - | \n",
+ " - | \n",
+ " id_7 | \n",
+ " id_8 | \n",
+ " id_9 | \n",
+ " - | \n",
+ "
\n",
+ " \n",
+ " | ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ "
\n",
+ " \n",
+ " | 2695 | \n",
+ " The Claudine was leaving next morning for Honolulu . | \n",
+ " 0020 | \n",
+ " {'model': 'm2', 'sent_id': 7, 'speaker': '0020', 'accent': None} | \n",
+ " {'model': 'm1', 'sent_id': 7, 'speaker': '0020', 'accent': None} | \n",
+ " {'model': 'm3', 'sent_id': 7, 'speaker': '0020', 'accent': None} | \n",
+ " The Claudine was leaving next morning for Honolulu . | \n",
+ "
\n",
+ " \n",
+ " | 2696 | \n",
+ " - | \n",
+ " - | \n",
+ " id_4045 | \n",
+ " id_4046 | \n",
+ " id_4047 | \n",
+ " - | \n",
+ "
\n",
+ " \n",
+ " | 2697 | \n",
+ " Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. | \n",
+ " 0020 | \n",
+ " {'model': 'm1', 'sent_id': 8, 'speaker': '0020', 'accent': None} | \n",
+ " {'model': 'm2', 'sent_id': 8, 'speaker': '0020', 'accent': None} | \n",
+ " {'model': 'm3', 'sent_id': 8, 'speaker': '0020', 'accent': None} | \n",
+ " Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. | \n",
+ "
\n",
+ " \n",
+ " | 2698 | \n",
+ " - | \n",
+ " - | \n",
+ " id_4048 | \n",
+ " id_4049 | \n",
+ " id_4050 | \n",
+ " - | \n",
+ "
\n",
+ " \n",
+ " | 2699 | \n",
+ " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ " 0020 | \n",
+ " {'model': 'm3', 'sent_id': 9, 'speaker': '0020', 'accent': None} | \n",
+ " {'model': 'm2', 'sent_id': 9, 'speaker': '0020', 'accent': None} | \n",
+ " {'model': 'm1', 'sent_id': 9, 'speaker': '0020', 'accent': None} | \n",
+ " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
2700 rows × 6 columns
\n",
+ "
\n",
+ "\n",
+ "
\n",
+ " \n",
+ " \n",
+ " | \n",
+ " Unnamed: 0 | \n",
+ " m1_text | \n",
+ " m1_accent_speaker | \n",
+ " m1 | \n",
+ " m2 | \n",
+ " m3 | \n",
+ " m2_text | \n",
+ "
\n",
+ " \n",
+ " \n",
+ " \n",
+ " | 0 | \n",
+ " 0 | \n",
+ " - | \n",
+ " - | \n",
+ " id_1 | \n",
+ " id_2 | \n",
+ " id_3 | \n",
+ " - | \n",
+ "
\n",
+ " \n",
+ " | 1 | \n",
+ " 1 | \n",
+ " We got few vegetables and fruits , and became fish eaters . | \n",
+ " 0011 | \n",
+ " 2 | \n",
+ " 2 | \n",
+ " 1 | \n",
+ " We got few vegetables and fruits , and became fish eaters . | \n",
+ "
\n",
+ " \n",
+ " | 2 | \n",
+ " 2 | \n",
+ " - | \n",
+ " - | \n",
+ " id_4 | \n",
+ " id_5 | \n",
+ " id_6 | \n",
+ " - | \n",
+ "
\n",
+ " \n",
+ " | 3 | \n",
+ " 3 | \n",
+ " Humans also judge distance by using the relative sizes of objects. | \n",
+ " 0011 | \n",
+ " 3 | \n",
+ " 3 | \n",
+ " 3 | \n",
+ " Humans also judge distance by using the relative sizes of objects. | \n",
+ "
\n",
+ " \n",
+ " | 4 | \n",
+ " 4 | \n",
+ " - | \n",
+ " - | \n",
+ " id_7 | \n",
+ " id_8 | \n",
+ " id_9 | \n",
+ " - | \n",
+ "
\n",
+ " \n",
+ " | ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ "
\n",
+ " \n",
+ " | 2695 | \n",
+ " 2695 | \n",
+ " The Claudine was leaving next morning for Honolulu . | \n",
+ " 0020 | \n",
+ " 2 | \n",
+ " 2 | \n",
+ " 3 | \n",
+ " The Claudine was leaving next morning for Honolulu . | \n",
+ "
\n",
+ " \n",
+ " | 2696 | \n",
+ " 2696 | \n",
+ " - | \n",
+ " - | \n",
+ " id_4045 | \n",
+ " id_4046 | \n",
+ " id_4047 | \n",
+ " - | \n",
+ "
\n",
+ " \n",
+ " | 2697 | \n",
+ " 2697 | \n",
+ " Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. | \n",
+ " 0020 | \n",
+ " 1 | \n",
+ " 2 | \n",
+ " 3 | \n",
+ " Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. | \n",
+ "
\n",
+ " \n",
+ " | 2698 | \n",
+ " 2698 | \n",
+ " - | \n",
+ " - | \n",
+ " id_4048 | \n",
+ " id_4049 | \n",
+ " id_4050 | \n",
+ " - | \n",
+ "
\n",
+ " \n",
+ " | 2699 | \n",
+ " 2699 | \n",
+ " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ " 0020 | \n",
+ " 1 | \n",
+ " 1 | \n",
+ " 2 | \n",
+ " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
2700 rows × 7 columns
\n",
+ "
\n",
+ "\n",
+ "
\n",
+ " \n",
+ " \n",
+ " | \n",
+ " model | \n",
+ " sent_id | \n",
+ " speaker | \n",
+ " accent | \n",
+ " text | \n",
+ " nmistakes | \n",
+ " pref_rank | \n",
+ " nwords | \n",
+ "
\n",
+ " \n",
+ " \n",
+ " \n",
+ " | id_3994 | \n",
+ " m2 | \n",
+ " 2 | \n",
+ " 0020 | \n",
+ " None | \n",
+ " Fifty yards ahead of her were the first of the rocks . | \n",
+ " 3 | \n",
+ " 3 | \n",
+ " 12 | \n",
+ "
\n",
+ " \n",
+ " | id_719 | \n",
+ " m1 | \n",
+ " 9 | \n",
+ " 0012 | \n",
+ " None | \n",
+ " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ " 3 | \n",
+ " 3 | \n",
+ " 11 | \n",
+ "
\n",
+ " \n",
+ " | id_4012 | \n",
+ " m3 | \n",
+ " 25 | \n",
+ " 0020 | \n",
+ " None | \n",
+ " Young people want to feel supported and appreciated by their company and their superiors. | \n",
+ " 2 | \n",
+ " 2 | \n",
+ " 14 | \n",
+ "
\n",
+ " \n",
+ " | id_3258 | \n",
+ " m3 | \n",
+ " 14 | \n",
+ " 0018 | \n",
+ " None | \n",
+ " When I reached Atlanta my steadily increasing disappointment was not lessened. I found it a big, dull, red town. | \n",
+ " 3 | \n",
+ " 3 | \n",
+ " 19 | \n",
+ "
\n",
+ " \n",
+ " | id_1331 | \n",
+ " m2 | \n",
+ " 30 | \n",
+ " 0013 | \n",
+ " None | \n",
+ " All this is thanks to his childhood in the mountains and to genetics, but it is his mental strength that sets him apart. | \n",
+ " 1 | \n",
+ " 1 | \n",
+ " 23 | \n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
\n",
+ "\n",
+ "
\n",
+ " \n",
+ " \n",
+ " | \n",
+ " model | \n",
+ " sent_id | \n",
+ " speaker | \n",
+ " accent | \n",
+ " text | \n",
+ " nmistakes | \n",
+ " pref_rank | \n",
+ " nwords | \n",
+ "
\n",
+ " \n",
+ " \n",
+ " \n",
+ " | id_3107 | \n",
+ " m1 | \n",
+ " 23 | \n",
+ " 0017 | \n",
+ " None | \n",
+ " For more than two hundred years the pessimists have been winning the public debate. | \n",
+ " FAILED | \n",
+ " FAILED | \n",
+ " 14 | \n",
+ "
\n",
+ " \n",
+ " | id_3112 | \n",
+ " m1 | \n",
+ " 25 | \n",
+ " 0017 | \n",
+ " None | \n",
+ " Young people want to feel supported and appreciated by their company and their superiors. | \n",
+ " FAILED | \n",
+ " FAILED | \n",
+ " 14 | \n",
+ "
\n",
+ " \n",
+ " | id_3108 | \n",
+ " m2 | \n",
+ " 23 | \n",
+ " 0017 | \n",
+ " None | \n",
+ " For more than two hundred years the pessimists have been winning the public debate. | \n",
+ " FAILED | \n",
+ " FAILED | \n",
+ " 14 | \n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
\n",
+ "\n",
+ "
\n",
+ " \n",
+ " \n",
+ " | \n",
+ " model | \n",
+ " sent_id | \n",
+ " speaker | \n",
+ " accent | \n",
+ " text | \n",
+ " nmistakes | \n",
+ " pref_rank | \n",
+ " nwords | \n",
+ "
\n",
+ " \n",
+ " \n",
+ " \n",
+ "
\n",
+ "
\n",
+ "\n",
+ "
\n",
+ " \n",
+ " \n",
+ " | \n",
+ " model | \n",
+ " sent_id | \n",
+ " speaker | \n",
+ " accent | \n",
+ " text | \n",
+ " nmistakes | \n",
+ " pref_rank | \n",
+ " nwords | \n",
+ " wer | \n",
+ "
\n",
+ " \n",
+ " \n",
+ " \n",
+ " | id_682 | \n",
+ " m2 | \n",
+ " 25 | \n",
+ " 0012 | \n",
+ " None | \n",
+ " Young people want to feel supported and appreciated by their company and their superiors. | \n",
+ " 1 | \n",
+ " 1 | \n",
+ " 14 | \n",
+ " 0.071429 | \n",
+ "
\n",
+ " \n",
+ " | id_1709 | \n",
+ " m2 | \n",
+ " 9 | \n",
+ " 0014 | \n",
+ " None | \n",
+ " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ " 2 | \n",
+ " 2 | \n",
+ " 11 | \n",
+ " 0.181818 | \n",
+ "
\n",
+ " \n",
+ " | id_3289 | \n",
+ " m2 | \n",
+ " 24 | \n",
+ " 0018 | \n",
+ " None | \n",
+ " It's wearing me out trying to juggle work with looking after my children and my family. | \n",
+ " 2 | \n",
+ " 2 | \n",
+ " 16 | \n",
+ " 0.125 | \n",
+ "
\n",
+ " \n",
+ " | id_3131 | \n",
+ " m1 | \n",
+ " 30 | \n",
+ " 0017 | \n",
+ " None | \n",
+ " All this is thanks to his childhood in the mountains and to genetics, but it is his mental strength that sets him apart. | \n",
+ " 2 | \n",
+ " 2 | \n",
+ " 23 | \n",
+ " 0.086957 | \n",
+ "
\n",
+ " \n",
+ " | id_807 | \n",
+ " m3 | \n",
+ " 8 | \n",
+ " 0012 | \n",
+ " None | \n",
+ " Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. | \n",
+ " 2 | \n",
+ " 2 | \n",
+ " 14 | \n",
+ " 0.142857 | \n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
\n",
+ "\n",
+ "
\n",
+ " \n",
+ " \n",
+ " | \n",
+ " mean | \n",
+ " std | \n",
+ " conf_int | \n",
+ " speaker_model | \n",
+ " speaker | \n",
+ "
\n",
+ " \n",
+ " \n",
+ " \n",
+ " | 0011_m1 | \n",
+ " 0.164942 | \n",
+ " 0.098481 | \n",
+ " (0.06646108959760694, 0.2634224229943425) | \n",
+ " 0011_m1 | \n",
+ " 0011 | \n",
+ "
\n",
+ " \n",
+ " | 0011_m2 | \n",
+ " 0.153745 | \n",
+ " 0.093724 | \n",
+ " (0.060021064505152905, 0.24746906731497884) | \n",
+ " 0011_m2 | \n",
+ " 0011 | \n",
+ "
\n",
+ " \n",
+ " | 0011_m3 | \n",
+ " 0.157573 | \n",
+ " 0.096471 | \n",
+ " (0.061102902232036266, 0.25404395873064856) | \n",
+ " 0011_m3 | \n",
+ " 0011 | \n",
+ "
\n",
+ " \n",
+ " | 0012_m1 | \n",
+ " 0.158086 | \n",
+ " 0.089081 | \n",
+ " (0.06900524445812664, 0.2471667344722284) | \n",
+ " 0012_m1 | \n",
+ " 0012 | \n",
+ "
\n",
+ " \n",
+ " | 0012_m2 | \n",
+ " 0.159536 | \n",
+ " 0.097171 | \n",
+ " (0.062365555572831854, 0.2567071476921302) | \n",
+ " 0012_m2 | \n",
+ " 0012 | \n",
+ "
\n",
+ " \n",
+ " | 0012_m3 | \n",
+ " 0.152597 | \n",
+ " 0.078147 | \n",
+ " (0.07444987547333512, 0.23074403505813826) | \n",
+ " 0012_m3 | \n",
+ " 0012 | \n",
+ "
\n",
+ " \n",
+ " | 0013_m1 | \n",
+ " 0.162392 | \n",
+ " 0.097572 | \n",
+ " (0.06482034517441204, 0.2599635226114294) | \n",
+ " 0013_m1 | \n",
+ " 0013 | \n",
+ "
\n",
+ " \n",
+ " | 0013_m2 | \n",
+ " 0.160631 | \n",
+ " 0.087547 | \n",
+ " (0.07308382549676895, 0.2481777627865256) | \n",
+ " 0013_m2 | \n",
+ " 0013 | \n",
+ "
\n",
+ " \n",
+ " | 0013_m3 | \n",
+ " 0.151915 | \n",
+ " 0.083490 | \n",
+ " (0.06842450305797412, 0.23540548757671007) | \n",
+ " 0013_m3 | \n",
+ " 0013 | \n",
+ "
\n",
+ " \n",
+ " | 0014_m1 | \n",
+ " 0.153306 | \n",
+ " 0.087367 | \n",
+ " (0.06593880323377925, 0.24067260289293907) | \n",
+ " 0014_m1 | \n",
+ " 0014 | \n",
+ "
\n",
+ " \n",
+ " | 0014_m2 | \n",
+ " 0.142629 | \n",
+ " 0.076339 | \n",
+ " (0.06629062112240179, 0.2189679924873874) | \n",
+ " 0014_m2 | \n",
+ " 0014 | \n",
+ "
\n",
+ " \n",
+ " | 0014_m3 | \n",
+ " 0.138416 | \n",
+ " 0.074690 | \n",
+ " (0.06372618801962594, 0.2131064901779155) | \n",
+ " 0014_m3 | \n",
+ " 0014 | \n",
+ "
\n",
+ " \n",
+ " | 0015_m1 | \n",
+ " 0.157652 | \n",
+ " 0.093588 | \n",
+ " (0.06406410579781929, 0.25123986956574174) | \n",
+ " 0015_m1 | \n",
+ " 0015 | \n",
+ "
\n",
+ " \n",
+ " | 0015_m2 | \n",
+ " 0.150783 | \n",
+ " 0.081496 | \n",
+ " (0.0692870412092566, 0.2322780794350478) | \n",
+ " 0015_m2 | \n",
+ " 0015 | \n",
+ "
\n",
+ " \n",
+ " | 0015_m3 | \n",
+ " 0.146370 | \n",
+ " 0.076419 | \n",
+ " (0.06995057276043903, 0.22278909244584022) | \n",
+ " 0015_m3 | \n",
+ " 0015 | \n",
+ "
\n",
+ " \n",
+ " | 0016_m1 | \n",
+ " 0.150378 | \n",
+ " 0.091428 | \n",
+ " (0.05895012298268368, 0.24180616747952347) | \n",
+ " 0016_m1 | \n",
+ " 0016 | \n",
+ "
\n",
+ " \n",
+ " | 0016_m2 | \n",
+ " 0.150525 | \n",
+ " 0.088010 | \n",
+ " (0.06251536658901205, 0.2385350291180889) | \n",
+ " 0016_m2 | \n",
+ " 0016 | \n",
+ "
\n",
+ " \n",
+ " | 0016_m3 | \n",
+ " 0.158251 | \n",
+ " 0.092847 | \n",
+ " (0.06540395567796967, 0.2510978352363106) | \n",
+ " 0016_m3 | \n",
+ " 0016 | \n",
+ "
\n",
+ " \n",
+ " | 0017_m1 | \n",
+ " 0.148413 | \n",
+ " 0.086607 | \n",
+ " (0.06180513203699889, 0.23501987280214826) | \n",
+ " 0017_m1 | \n",
+ " 0017 | \n",
+ "
\n",
+ " \n",
+ " | 0017_m2 | \n",
+ " 0.161314 | \n",
+ " 0.086679 | \n",
+ " (0.07463489425964775, 0.24799248779959487) | \n",
+ " 0017_m2 | \n",
+ " 0017 | \n",
+ "
\n",
+ " \n",
+ " | 0017_m3 | \n",
+ " 0.143036 | \n",
+ " 0.077084 | \n",
+ " (0.06595177268079003, 0.22011948254843597) | \n",
+ " 0017_m3 | \n",
+ " 0017 | \n",
+ "
\n",
+ " \n",
+ " | 0018_m1 | \n",
+ " 0.153057 | \n",
+ " 0.087590 | \n",
+ " (0.06546767387862472, 0.2406472593275521) | \n",
+ " 0018_m1 | \n",
+ " 0018 | \n",
+ "
\n",
+ " \n",
+ " | 0018_m2 | \n",
+ " 0.152221 | \n",
+ " 0.086864 | \n",
+ " (0.06535729973008131, 0.23908491722641328) | \n",
+ " 0018_m2 | \n",
+ " 0018 | \n",
+ "
\n",
+ " \n",
+ " | 0018_m3 | \n",
+ " 0.149037 | \n",
+ " 0.079778 | \n",
+ " (0.06925902808914024, 0.22881432417513253) | \n",
+ " 0018_m3 | \n",
+ " 0018 | \n",
+ "
\n",
+ " \n",
+ " | 0020_m1 | \n",
+ " 0.154046 | \n",
+ " 0.094158 | \n",
+ " (0.05988797395800191, 0.24820440171417577) | \n",
+ " 0020_m1 | \n",
+ " 0020 | \n",
+ "
\n",
+ " \n",
+ " | 0020_m2 | \n",
+ " 0.164588 | \n",
+ " 0.091769 | \n",
+ " (0.07281827754800148, 0.25635701075077355) | \n",
+ " 0020_m2 | \n",
+ " 0020 | \n",
+ "
\n",
+ " \n",
+ " | 0020_m3 | \n",
+ " 0.158046 | \n",
+ " 0.091050 | \n",
+ " (0.06699555516117969, 0.24909578894879178) | \n",
+ " 0020_m3 | \n",
+ " 0020 | \n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
\n",
" | 1 | \n",
- " generated_hifi/esd_tune/g_3164999/0011/angry/1.wav | \n",
- " generated_hifi/esd_tune_reversal/g_3164999/0011/angry/1.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_3164999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune/g_2669999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_2669999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/1.wav | \n",
" 1.wav | \n",
" 0011 | \n",
- " We got few vegetables and fruits , and became fish eaters . | \n",
+ " Do you realize what time it is? | \n",
" 1.wav | \n",
" 0011 | \n",
- " We got few vegetables and fruits , and became fish eaters . | \n",
+ " Do you realize what time it is? | \n",
" 1.wav | \n",
" 0011 | \n",
- " We got few vegetables and fruits , and became fish eaters . | \n",
+ " Do you realize what time it is? | \n",
"
\n",
" \n",
" | 2 | \n",
@@ -537,18 +536,18 @@
"
\n",
" \n",
" | 3 | \n",
- " generated_hifi/esd_tune/g_3164999/0011/angry/10.wav | \n",
- " generated_hifi/esd_tune_reversal/g_3164999/0011/angry/10.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_3164999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune/g_2669999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_2669999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/10.wav | \n",
" 10.wav | \n",
" 0011 | \n",
- " Humans also judge distance by using the relative sizes of objects. | \n",
+ " What are you still doing here? | \n",
" 10.wav | \n",
" 0011 | \n",
- " Humans also judge distance by using the relative sizes of objects. | \n",
+ " What are you still doing here? | \n",
" 10.wav | \n",
" 0011 | \n",
- " Humans also judge distance by using the relative sizes of objects. | \n",
+ " What are you still doing here? | \n",
"
\n",
" \n",
" | 4 | \n",
@@ -582,18 +581,18 @@
"
\n",
" \n",
" | 2695 | \n",
- " generated_hifi/esd_tune/g_3164999/0020/surprise/7.wav | \n",
- " generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/7.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune/g_2669999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/7.wav | \n",
" 7.wav | \n",
" 0020 | \n",
- " The Claudine was leaving next morning for Honolulu . | \n",
+ " Only one vehicle may be allowed to park at any given time. | \n",
" 7.wav | \n",
" 0020 | \n",
- " The Claudine was leaving next morning for Honolulu . | \n",
+ " Only one vehicle may be allowed to park at any given time. | \n",
" 7.wav | \n",
" 0020 | \n",
- " The Claudine was leaving next morning for Honolulu . | \n",
+ " Only one vehicle may be allowed to park at any given time. | \n",
"
\n",
" \n",
" | 2696 | \n",
@@ -612,18 +611,18 @@
"
\n",
" \n",
" | 2697 | \n",
- " generated_hifi/esd_tune/g_3164999/0020/surprise/8.wav | \n",
- " generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/8.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune/g_2669999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/8.wav | \n",
" 8.wav | \n",
" 0020 | \n",
- " Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. | \n",
+ " The deadlines are indeed very tight. | \n",
" 8.wav | \n",
" 0020 | \n",
- " Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. | \n",
+ " The deadlines are indeed very tight. | \n",
" 8.wav | \n",
" 0020 | \n",
- " Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. | \n",
+ " The deadlines are indeed very tight. | \n",
"
\n",
" \n",
" | 2698 | \n",
@@ -642,18 +641,18 @@
"
\n",
" \n",
" | 2699 | \n",
- " generated_hifi/esd_tune/g_3164999/0020/surprise/9.wav | \n",
- " generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/9.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune/g_2669999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/9.wav | \n",
" 9.wav | \n",
" 0020 | \n",
- " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ " I'm glad you enjoyed yourself. | \n",
" 9.wav | \n",
" 0020 | \n",
- " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ " I'm glad you enjoyed yourself. | \n",
" 9.wav | \n",
" 0020 | \n",
- " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ " I'm glad you enjoyed yourself. | \n",
"
\n",
" \n",
"\n",
@@ -663,42 +662,42 @@
"text/plain": [
" m1 \\\n",
"0 id_1 \n",
- "1 generated_hifi/esd_tune/g_3164999/0011/angry/1.wav \n",
+ "1 generated_hifi/esd_tune/g_2669999/0011/angry/1.wav \n",
"2 id_4 \n",
- "3 generated_hifi/esd_tune/g_3164999/0011/angry/10.wav \n",
+ "3 generated_hifi/esd_tune/g_2669999/0011/angry/10.wav \n",
"4 id_7 \n",
"... ... \n",
- "2695 generated_hifi/esd_tune/g_3164999/0020/surprise/7.wav \n",
+ "2695 generated_hifi/esd_tune/g_2669999/0020/surprise/7.wav \n",
"2696 id_4045 \n",
- "2697 generated_hifi/esd_tune/g_3164999/0020/surprise/8.wav \n",
+ "2697 generated_hifi/esd_tune/g_2669999/0020/surprise/8.wav \n",
"2698 id_4048 \n",
- "2699 generated_hifi/esd_tune/g_3164999/0020/surprise/9.wav \n",
+ "2699 generated_hifi/esd_tune/g_2669999/0020/surprise/9.wav \n",
"\n",
" m2 \\\n",
"0 id_2 \n",
- "1 generated_hifi/esd_tune_reversal/g_3164999/0011/angry/1.wav \n",
+ "1 generated_hifi/esd_tune_reversal/g_2669999/0011/angry/1.wav \n",
"2 id_5 \n",
- "3 generated_hifi/esd_tune_reversal/g_3164999/0011/angry/10.wav \n",
+ "3 generated_hifi/esd_tune_reversal/g_2669999/0011/angry/10.wav \n",
"4 id_8 \n",
"... ... \n",
- "2695 generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/7.wav \n",
+ "2695 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/7.wav \n",
"2696 id_4046 \n",
- "2697 generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/8.wav \n",
+ "2697 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/8.wav \n",
"2698 id_4049 \n",
- "2699 generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/9.wav \n",
+ "2699 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/9.wav \n",
"\n",
" m3 \\\n",
"0 id_3 \n",
- "1 generated_hifi/esd_tune_advloss0/g_3164999/0011/angry/1.wav \n",
+ "1 generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/1.wav \n",
"2 id_6 \n",
- "3 generated_hifi/esd_tune_advloss0/g_3164999/0011/angry/10.wav \n",
+ "3 generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/10.wav \n",
"4 id_9 \n",
"... ... \n",
- "2695 generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/7.wav \n",
+ "2695 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/7.wav \n",
"2696 id_4047 \n",
- "2697 generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/8.wav \n",
+ "2697 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/8.wav \n",
"2698 id_4050 \n",
- "2699 generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/9.wav \n",
+ "2699 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/9.wav \n",
"\n",
" m1_wavbasename m1_accent_speaker \\\n",
"0 None None \n",
@@ -713,18 +712,18 @@
"2698 None None \n",
"2699 9.wav 0020 \n",
"\n",
- " m1_text \\\n",
- "0 None \n",
- "1 We got few vegetables and fruits , and became fish eaters . \n",
- "2 None \n",
- "3 Humans also judge distance by using the relative sizes of objects. \n",
- "4 None \n",
- "... ... \n",
- "2695 The Claudine was leaving next morning for Honolulu . \n",
- "2696 None \n",
- "2697 Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. \n",
- "2698 None \n",
- "2699 Different telescope designs perform differently, and have different strengths and weaknesses. \n",
+ " m1_text \\\n",
+ "0 None \n",
+ "1 Do you realize what time it is? \n",
+ "2 None \n",
+ "3 What are you still doing here? \n",
+ "4 None \n",
+ "... ... \n",
+ "2695 Only one vehicle may be allowed to park at any given time. \n",
+ "2696 None \n",
+ "2697 The deadlines are indeed very tight. \n",
+ "2698 None \n",
+ "2699 I'm glad you enjoyed yourself. \n",
"\n",
" m2_wavbasename m2_accent_speaker \\\n",
"0 None None \n",
@@ -739,18 +738,18 @@
"2698 None None \n",
"2699 9.wav 0020 \n",
"\n",
- " m2_text \\\n",
- "0 None \n",
- "1 We got few vegetables and fruits , and became fish eaters . \n",
- "2 None \n",
- "3 Humans also judge distance by using the relative sizes of objects. \n",
- "4 None \n",
- "... ... \n",
- "2695 The Claudine was leaving next morning for Honolulu . \n",
- "2696 None \n",
- "2697 Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. \n",
- "2698 None \n",
- "2699 Different telescope designs perform differently, and have different strengths and weaknesses. \n",
+ " m2_text \\\n",
+ "0 None \n",
+ "1 Do you realize what time it is? \n",
+ "2 None \n",
+ "3 What are you still doing here? \n",
+ "4 None \n",
+ "... ... \n",
+ "2695 Only one vehicle may be allowed to park at any given time. \n",
+ "2696 None \n",
+ "2697 The deadlines are indeed very tight. \n",
+ "2698 None \n",
+ "2699 I'm glad you enjoyed yourself. \n",
"\n",
" m3_wavbasename m3_accent_speaker \\\n",
"0 None None \n",
@@ -765,18 +764,18 @@
"2698 None None \n",
"2699 9.wav 0020 \n",
"\n",
- " m3_text \n",
- "0 None \n",
- "1 We got few vegetables and fruits , and became fish eaters . \n",
- "2 None \n",
- "3 Humans also judge distance by using the relative sizes of objects. \n",
- "4 None \n",
- "... ... \n",
- "2695 The Claudine was leaving next morning for Honolulu . \n",
- "2696 None \n",
- "2697 Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. \n",
- "2698 None \n",
- "2699 Different telescope designs perform differently, and have different strengths and weaknesses. \n",
+ " m3_text \n",
+ "0 None \n",
+ "1 Do you realize what time it is? \n",
+ "2 None \n",
+ "3 What are you still doing here? \n",
+ "4 None \n",
+ "... ... \n",
+ "2695 Only one vehicle may be allowed to park at any given time. \n",
+ "2696 None \n",
+ "2697 The deadlines are indeed very tight. \n",
+ "2698 None \n",
+ "2699 I'm glad you enjoyed yourself. \n",
"\n",
"[2700 rows x 12 columns]"
]
@@ -925,12 +924,12 @@
" \n",
" \n",
" | 1 | \n",
- " We got few vegetables and fruits , and became fish eaters . | \n",
+ " Do you realize what time it is? | \n",
" 0011 | \n",
- " generated_hifi/esd_tune/g_3164999/0011/angry/1.wav | \n",
- " generated_hifi/esd_tune_reversal/g_3164999/0011/angry/1.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_3164999/0011/angry/1.wav | \n",
- " We got few vegetables and fruits , and became fish eaters . | \n",
+ " generated_hifi/esd_tune/g_2669999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_2669999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/1.wav | \n",
+ " Do you realize what time it is? | \n",
"
\n",
" \n",
" | 2 | \n",
@@ -943,12 +942,12 @@
"
\n",
" \n",
" | 3 | \n",
- " Humans also judge distance by using the relative sizes of objects. | \n",
+ " What are you still doing here? | \n",
" 0011 | \n",
- " generated_hifi/esd_tune/g_3164999/0011/angry/10.wav | \n",
- " generated_hifi/esd_tune_reversal/g_3164999/0011/angry/10.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_3164999/0011/angry/10.wav | \n",
- " Humans also judge distance by using the relative sizes of objects. | \n",
+ " generated_hifi/esd_tune/g_2669999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_2669999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/10.wav | \n",
+ " What are you still doing here? | \n",
"
\n",
" \n",
" | 4 | \n",
@@ -970,12 +969,12 @@
"
\n",
" \n",
" | 2695 | \n",
- " The Claudine was leaving next morning for Honolulu . | \n",
+ " Only one vehicle may be allowed to park at any given time. | \n",
" 0020 | \n",
- " generated_hifi/esd_tune/g_3164999/0020/surprise/7.wav | \n",
- " generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/7.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/7.wav | \n",
- " The Claudine was leaving next morning for Honolulu . | \n",
+ " generated_hifi/esd_tune/g_2669999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/7.wav | \n",
+ " Only one vehicle may be allowed to park at any given time. | \n",
"
\n",
" \n",
" | 2696 | \n",
@@ -988,12 +987,12 @@
"
\n",
" \n",
" | 2697 | \n",
- " Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. | \n",
+ " The deadlines are indeed very tight. | \n",
" 0020 | \n",
- " generated_hifi/esd_tune/g_3164999/0020/surprise/8.wav | \n",
- " generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/8.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/8.wav | \n",
- " Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. | \n",
+ " generated_hifi/esd_tune/g_2669999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/8.wav | \n",
+ " The deadlines are indeed very tight. | \n",
"
\n",
" \n",
" | 2698 | \n",
@@ -1006,12 +1005,12 @@
"
\n",
" \n",
" | 2699 | \n",
- " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ " I'm glad you enjoyed yourself. | \n",
" 0020 | \n",
- " generated_hifi/esd_tune/g_3164999/0020/surprise/9.wav | \n",
- " generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/9.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/9.wav | \n",
- " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ " generated_hifi/esd_tune/g_2669999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/9.wav | \n",
+ " I'm glad you enjoyed yourself. | \n",
"
\n",
" \n",
"\n",
@@ -1019,70 +1018,70 @@
""
],
"text/plain": [
- " m1_text \\\n",
- "0 None \n",
- "1 We got few vegetables and fruits , and became fish eaters . \n",
- "2 None \n",
- "3 Humans also judge distance by using the relative sizes of objects. \n",
- "4 None \n",
- "... ... \n",
- "2695 The Claudine was leaving next morning for Honolulu . \n",
- "2696 None \n",
- "2697 Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. \n",
- "2698 None \n",
- "2699 Different telescope designs perform differently, and have different strengths and weaknesses. \n",
+ " m1_text \\\n",
+ "0 None \n",
+ "1 Do you realize what time it is? \n",
+ "2 None \n",
+ "3 What are you still doing here? \n",
+ "4 None \n",
+ "... ... \n",
+ "2695 Only one vehicle may be allowed to park at any given time. \n",
+ "2696 None \n",
+ "2697 The deadlines are indeed very tight. \n",
+ "2698 None \n",
+ "2699 I'm glad you enjoyed yourself. \n",
"\n",
" m1_accent_speaker m1 \\\n",
"0 None id_1 \n",
- "1 0011 generated_hifi/esd_tune/g_3164999/0011/angry/1.wav \n",
+ "1 0011 generated_hifi/esd_tune/g_2669999/0011/angry/1.wav \n",
"2 None id_4 \n",
- "3 0011 generated_hifi/esd_tune/g_3164999/0011/angry/10.wav \n",
+ "3 0011 generated_hifi/esd_tune/g_2669999/0011/angry/10.wav \n",
"4 None id_7 \n",
"... ... ... \n",
- "2695 0020 generated_hifi/esd_tune/g_3164999/0020/surprise/7.wav \n",
+ "2695 0020 generated_hifi/esd_tune/g_2669999/0020/surprise/7.wav \n",
"2696 None id_4045 \n",
- "2697 0020 generated_hifi/esd_tune/g_3164999/0020/surprise/8.wav \n",
+ "2697 0020 generated_hifi/esd_tune/g_2669999/0020/surprise/8.wav \n",
"2698 None id_4048 \n",
- "2699 0020 generated_hifi/esd_tune/g_3164999/0020/surprise/9.wav \n",
+ "2699 0020 generated_hifi/esd_tune/g_2669999/0020/surprise/9.wav \n",
"\n",
" m2 \\\n",
"0 id_2 \n",
- "1 generated_hifi/esd_tune_reversal/g_3164999/0011/angry/1.wav \n",
+ "1 generated_hifi/esd_tune_reversal/g_2669999/0011/angry/1.wav \n",
"2 id_5 \n",
- "3 generated_hifi/esd_tune_reversal/g_3164999/0011/angry/10.wav \n",
+ "3 generated_hifi/esd_tune_reversal/g_2669999/0011/angry/10.wav \n",
"4 id_8 \n",
"... ... \n",
- "2695 generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/7.wav \n",
+ "2695 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/7.wav \n",
"2696 id_4046 \n",
- "2697 generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/8.wav \n",
+ "2697 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/8.wav \n",
"2698 id_4049 \n",
- "2699 generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/9.wav \n",
+ "2699 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/9.wav \n",
"\n",
" m3 \\\n",
"0 id_3 \n",
- "1 generated_hifi/esd_tune_advloss0/g_3164999/0011/angry/1.wav \n",
+ "1 generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/1.wav \n",
"2 id_6 \n",
- "3 generated_hifi/esd_tune_advloss0/g_3164999/0011/angry/10.wav \n",
+ "3 generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/10.wav \n",
"4 id_9 \n",
"... ... \n",
- "2695 generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/7.wav \n",
+ "2695 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/7.wav \n",
"2696 id_4047 \n",
- "2697 generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/8.wav \n",
+ "2697 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/8.wav \n",
"2698 id_4050 \n",
- "2699 generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/9.wav \n",
+ "2699 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/9.wav \n",
"\n",
- " m2_text \n",
- "0 None \n",
- "1 We got few vegetables and fruits , and became fish eaters . \n",
- "2 None \n",
- "3 Humans also judge distance by using the relative sizes of objects. \n",
- "4 None \n",
- "... ... \n",
- "2695 The Claudine was leaving next morning for Honolulu . \n",
- "2696 None \n",
- "2697 Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. \n",
- "2698 None \n",
- "2699 Different telescope designs perform differently, and have different strengths and weaknesses. \n",
+ " m2_text \n",
+ "0 None \n",
+ "1 Do you realize what time it is? \n",
+ "2 None \n",
+ "3 What are you still doing here? \n",
+ "4 None \n",
+ "... ... \n",
+ "2695 Only one vehicle may be allowed to park at any given time. \n",
+ "2696 None \n",
+ "2697 The deadlines are indeed very tight. \n",
+ "2698 None \n",
+ "2699 I'm glad you enjoyed yourself. \n",
"\n",
"[2700 rows x 6 columns]"
]
@@ -1150,12 +1149,12 @@
" \n",
" \n",
" | 1 | \n",
- " We got few vegetables and fruits , and became fish eaters . | \n",
+ " Do you realize what time it is? | \n",
" 0011 | \n",
- " generated_hifi/esd_tune_reversal/g_3164999/0011/angry/1.wav | \n",
- " generated_hifi/esd_tune/g_3164999/0011/angry/1.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_3164999/0011/angry/1.wav | \n",
- " We got few vegetables and fruits , and became fish eaters . | \n",
+ " generated_hifi/esd_tune_reversal/g_2669999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune/g_2669999/0011/angry/1.wav | \n",
+ " Do you realize what time it is? | \n",
"
\n",
" \n",
" | 2 | \n",
@@ -1168,12 +1167,12 @@
"
\n",
" \n",
" | 3 | \n",
- " Humans also judge distance by using the relative sizes of objects. | \n",
+ " What are you still doing here? | \n",
" 0011 | \n",
- " generated_hifi/esd_tune_reversal/g_3164999/0011/angry/10.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_3164999/0011/angry/10.wav | \n",
- " generated_hifi/esd_tune/g_3164999/0011/angry/10.wav | \n",
- " Humans also judge distance by using the relative sizes of objects. | \n",
+ " generated_hifi/esd_tune/g_2669999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_2669999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/10.wav | \n",
+ " What are you still doing here? | \n",
"
\n",
" \n",
" | 4 | \n",
@@ -1195,12 +1194,12 @@
"
\n",
" \n",
" | 2695 | \n",
- " The Claudine was leaving next morning for Honolulu . | \n",
+ " Only one vehicle may be allowed to park at any given time. | \n",
" 0020 | \n",
- " generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/7.wav | \n",
- " generated_hifi/esd_tune/g_3164999/0020/surprise/7.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/7.wav | \n",
- " The Claudine was leaving next morning for Honolulu . | \n",
+ " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune/g_2669999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/7.wav | \n",
+ " Only one vehicle may be allowed to park at any given time. | \n",
"
\n",
" \n",
" | 2696 | \n",
@@ -1213,12 +1212,12 @@
"
\n",
" \n",
" | 2697 | \n",
- " Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. | \n",
+ " The deadlines are indeed very tight. | \n",
" 0020 | \n",
- " generated_hifi/esd_tune/g_3164999/0020/surprise/8.wav | \n",
- " generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/8.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/8.wav | \n",
- " Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. | \n",
+ " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune/g_2669999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/8.wav | \n",
+ " The deadlines are indeed very tight. | \n",
"
\n",
" \n",
" | 2698 | \n",
@@ -1231,12 +1230,12 @@
"
\n",
" \n",
" | 2699 | \n",
- " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ " I'm glad you enjoyed yourself. | \n",
" 0020 | \n",
- " generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/9.wav | \n",
- " generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/9.wav | \n",
- " generated_hifi/esd_tune/g_3164999/0020/surprise/9.wav | \n",
- " Different telescope designs perform differently, and have different strengths and weaknesses. | \n",
+ " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune/g_2669999/0020/surprise/9.wav | \n",
+ " I'm glad you enjoyed yourself. | \n",
"
\n",
" \n",
"\n",
@@ -1244,18 +1243,18 @@
""
],
"text/plain": [
- " m1_text \\\n",
- "0 - \n",
- "1 We got few vegetables and fruits , and became fish eaters . \n",
- "2 - \n",
- "3 Humans also judge distance by using the relative sizes of objects. \n",
- "4 - \n",
- "... ... \n",
- "2695 The Claudine was leaving next morning for Honolulu . \n",
- "2696 - \n",
- "2697 Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. \n",
- "2698 - \n",
- "2699 Different telescope designs perform differently, and have different strengths and weaknesses. \n",
+ " m1_text \\\n",
+ "0 - \n",
+ "1 Do you realize what time it is? \n",
+ "2 - \n",
+ "3 What are you still doing here? \n",
+ "4 - \n",
+ "... ... \n",
+ "2695 Only one vehicle may be allowed to park at any given time. \n",
+ "2696 - \n",
+ "2697 The deadlines are indeed very tight. \n",
+ "2698 - \n",
+ "2699 I'm glad you enjoyed yourself. \n",
"\n",
" m1_accent_speaker \\\n",
"0 - \n",
@@ -1272,55 +1271,55 @@
"\n",
" m1 \\\n",
"0 id_1 \n",
- "1 generated_hifi/esd_tune_reversal/g_3164999/0011/angry/1.wav \n",
+ "1 generated_hifi/esd_tune_reversal/g_2669999/0011/angry/1.wav \n",
"2 id_4 \n",
- "3 generated_hifi/esd_tune_reversal/g_3164999/0011/angry/10.wav \n",
+ "3 generated_hifi/esd_tune/g_2669999/0011/angry/10.wav \n",
"4 id_7 \n",
"... ... \n",
- "2695 generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/7.wav \n",
+ "2695 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/7.wav \n",
"2696 id_4045 \n",
- "2697 generated_hifi/esd_tune/g_3164999/0020/surprise/8.wav \n",
+ "2697 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/8.wav \n",
"2698 id_4048 \n",
- "2699 generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/9.wav \n",
+ "2699 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/9.wav \n",
"\n",
" m2 \\\n",
"0 id_2 \n",
- "1 generated_hifi/esd_tune/g_3164999/0011/angry/1.wav \n",
+ "1 generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/1.wav \n",
"2 id_5 \n",
- "3 generated_hifi/esd_tune_advloss0/g_3164999/0011/angry/10.wav \n",
+ "3 generated_hifi/esd_tune_reversal/g_2669999/0011/angry/10.wav \n",
"4 id_8 \n",
"... ... \n",
- "2695 generated_hifi/esd_tune/g_3164999/0020/surprise/7.wav \n",
+ "2695 generated_hifi/esd_tune/g_2669999/0020/surprise/7.wav \n",
"2696 id_4046 \n",
- "2697 generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/8.wav \n",
+ "2697 generated_hifi/esd_tune/g_2669999/0020/surprise/8.wav \n",
"2698 id_4049 \n",
- "2699 generated_hifi/esd_tune_reversal/g_3164999/0020/surprise/9.wav \n",
+ "2699 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/9.wav \n",
"\n",
" m3 \\\n",
"0 id_3 \n",
- "1 generated_hifi/esd_tune_advloss0/g_3164999/0011/angry/1.wav \n",
+ "1 generated_hifi/esd_tune/g_2669999/0011/angry/1.wav \n",
"2 id_6 \n",
- "3 generated_hifi/esd_tune/g_3164999/0011/angry/10.wav \n",
+ "3 generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/10.wav \n",
"4 id_9 \n",
"... ... \n",
- "2695 generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/7.wav \n",
+ "2695 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/7.wav \n",
"2696 id_4047 \n",
- "2697 generated_hifi/esd_tune_advloss0/g_3164999/0020/surprise/8.wav \n",
+ "2697 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/8.wav \n",
"2698 id_4050 \n",
- "2699 generated_hifi/esd_tune/g_3164999/0020/surprise/9.wav \n",
+ "2699 generated_hifi/esd_tune/g_2669999/0020/surprise/9.wav \n",
"\n",
- " m2_text \n",
- "0 - \n",
- "1 We got few vegetables and fruits , and became fish eaters . \n",
- "2 - \n",
- "3 Humans also judge distance by using the relative sizes of objects. \n",
- "4 - \n",
- "... ... \n",
- "2695 The Claudine was leaving next morning for Honolulu . \n",
- "2696 - \n",
- "2697 Prosecutors have opened a massive investigation into allegations of fixing games and illegal betting. \n",
- "2698 - \n",
- "2699 Different telescope designs perform differently, and have different strengths and weaknesses. \n",
+ " m2_text \n",
+ "0 - \n",
+ "1 Do you realize what time it is? \n",
+ "2 - \n",
+ "3 What are you still doing here? \n",
+ "4 - \n",
+ "... ... \n",
+ "2695 Only one vehicle may be allowed to park at any given time. \n",
+ "2696 - \n",
+ "2697 The deadlines are indeed very tight. \n",
+ "2698 - \n",
+ "2699 I'm glad you enjoyed yourself. \n",
"\n",
"[2700 rows x 6 columns]"
]
@@ -1347,11 +1346,12 @@
},
{
"cell_type": "code",
- "execution_count": 14,
+ "execution_count": 15,
"id": "eef3b325-c26e-457a-b575-eb6b210aff7f",
"metadata": {},
"outputs": [],
"source": [
+ "Path('reports').mkdir(exist_ok=True)\n",
"df_permuted.to_csv(\n",
" 'reports/eval_map.csv',\n",
" sep = \"\\t\",\n",
@@ -1363,7 +1363,7 @@
},
{
"cell_type": "code",
- "execution_count": 15,
+ "execution_count": 16,
"id": "d32630c2-aca7-4ee9-b44c-a949d0fd17c4",
"metadata": {},
"outputs": [],
@@ -1380,7 +1380,7 @@
},
{
"cell_type": "code",
- "execution_count": 16,
+ "execution_count": 17,
"id": "2c11c3ca-abe4-4bbd-930f-8a4f62c9dc8f",
"metadata": {},
"outputs": [],
@@ -1403,7 +1403,7 @@
},
{
"cell_type": "code",
- "execution_count": 17,
+ "execution_count": 18,
"id": "11d84c76-841a-4215-a45f-268076e4ae21",
"metadata": {},
"outputs": [],
@@ -1415,7 +1415,7 @@
},
{
"cell_type": "code",
- "execution_count": 18,
+ "execution_count": 19,
"id": "26b577fe-3957-4902-8758-e5501f5d81a0",
"metadata": {},
"outputs": [],
@@ -1426,7 +1426,7 @@
},
{
"cell_type": "code",
- "execution_count": 19,
+ "execution_count": 20,
"id": "f37b7fb3-93eb-448c-a8a3-a23cb0d165ae",
"metadata": {},
"outputs": [],
@@ -1437,7 +1437,7 @@
},
{
"cell_type": "code",
- "execution_count": 20,
+ "execution_count": 21,
"id": "24aef771-215a-4fe6-82cb-1ff375681189",
"metadata": {
"scrolled": true,
@@ -1530,7 +1530,7 @@
},
{
"cell_type": "code",
- "execution_count": 21,
+ "execution_count": 22,
"id": "49a60285-31bf-4263-9162-c3d928aa9068",
"metadata": {},
"outputs": [
From 8b4a95e7a90f3845ec6715cfb956c4173297c9b7 Mon Sep 17 00:00:00 2001
From: ruk0sh <43880105+ruk0sh@users.noreply.github.com>
Date: Wed, 18 May 2022 18:51:44 +0300
Subject: [PATCH 16/17] Update test phrases
---
generate_listener_html.ipynb | 173 ++++++++++++++++++-----------------
1 file changed, 87 insertions(+), 86 deletions(-)
diff --git a/generate_listener_html.ipynb b/generate_listener_html.ipynb
index e6404dd..a23ed4d 100644
--- a/generate_listener_html.ipynb
+++ b/generate_listener_html.ipynb
@@ -28,35 +28,36 @@
"outputs": [],
"source": [
"texts = [\n",
- " 'Do you realize what time it is?',\n",
- " 'He comes back to the valley.',\n",
- " 'This dress does not look worth much!',\n",
- " 'What happened tonight has nothing to do with Henry.',\n",
- " 'Today, five years later, we are facing a similar situation.',\n",
- " 'When I saw you kissing, you looked really happy.',\n",
- " 'Only one vehicle may be allowed to park at any given time.',\n",
- " 'The deadlines are indeed very tight.',\n",
- " \"I'm glad you enjoyed yourself.\", 'What are you still doing here?',\n",
- " 'This is an animal that is admired for its whiteness and cleanliness. ',\n",
- " 'Perhaps there is another way to pose these issues.',\n",
- " \"Your students' test scores drop lower and lower every year.\",\n",
- " 'Wherever her tears fell, a fruit tree grew.',\n",
- " 'I was about to head back to my hotel and go to sleep.',\n",
- " 'You said she really helped last time.',\n",
- " 'My favorite season, spring, is here.',\n",
- " \"He's the rich guy who built the airplanes.\",\n",
- " 'Otto and Elizabeth gave it to us, for the wedding - incredibly generous.',\n",
- " 'Look, the police said that there was nothing stolen from the house.',\n",
- " 'And I suppose we can thank your brother for that.',\n",
- " \"That's a pretty dangerous thing you're doing.\",\n",
- " 'He arrived in Japan for the first time at the age of twenty six.',\n",
- " 'Sam thought we were having fun being together.',\n",
- " \"Well, the true value of something isn't always determined by its price.\",\n",
- " \"No, it's not polite to discuss a lady's age.\",\n",
- " \"Just another quarter-mile and I don't have to be tolerant ever again.\",\n",
- " \"But Jones' apartment had only been rented out for a week.\",\n",
- " 'What your perfect day would have been like?',\n",
- " 'Not a very useful skill, especially when the money runs out.',\n",
+ " 'Do you realize what time it is?',\n",
+ " 'He comes back to the valley.',\n",
+ " 'This dress does not look worth much!',\n",
+ " 'What happened tonight has nothing to do with Henry.',\n",
+ " 'Today, five years later, we are facing a similar situation.',\n",
+ " 'When I saw you kissing, you looked really happy.',\n",
+ " 'Only one vehicle may be allowed to park at any given time.',\n",
+ " 'The deadlines are indeed very tight.',\n",
+ " \"I'm glad you enjoyed yourself.\",\n",
+ " 'What are you still doing here?',\n",
+ " 'This is an animal that is admired for its whiteness and cleanliness. ',\n",
+ " 'Perhaps there is another way to pose these issues.',\n",
+ " \"Your students' test scores drop lower and lower every year.\",\n",
+ " 'Wherever her tears fell, a fruit tree grew.',\n",
+ " 'I was about to head back to my hotel and go to sleep.',\n",
+ " 'You said she really helped last time.',\n",
+ " 'My favorite season, spring, is here.',\n",
+ " \"He's the rich guy who built the airplanes.\",\n",
+ " 'Otto and Elizabeth gave it to us, for the wedding - incredibly generous.',\n",
+ " 'Look, the police said that there was nothing stolen from the house.',\n",
+ " 'And I suppose we can thank your brother for that.',\n",
+ " \"That's a pretty dangerous thing you're doing.\",\n",
+ " 'He arrived in Japan for the first time at the age of twenty six.',\n",
+ " 'Sam thought we were having fun being together.',\n",
+ " \"Well, the true value of something isn't always determined by its price.\",\n",
+ " \"No, it's not polite to discuss a lady's age.\",\n",
+ " \"Just another quarter-mile and I don't have to be tolerant ever again.\",\n",
+ " \"But Jones' apartment had only been rented out for a week.\",\n",
+ " 'What your perfect day would have been like?',\n",
+ " 'Not a very useful skill, especially when the money runs out.',\n",
"]"
]
},
@@ -68,36 +69,36 @@
"outputs": [],
"source": [
"huawei_phones = [\n",
- " 'D UW1 Y UW1 R IY1 AH0 L AY2 Z W AH1 T T AY1 M IH1 T IH1 Z ',\n",
- " 'HH IY1 K AH1 M Z B AE1 K T UW1 DH AH0 V AE1 L IY0 ',\n",
- " 'DH IH1 S D R EH1 S D AH1 Z N AA1 T L UH1 K W ER1 TH M AH1 CH ',\n",
- " 'W AH1 T HH AE1 P AH0 N D T AH0 N AY1 T HH AE1 Z N AH1 TH IH0 NG T UW1 D UW1 W IH1 DH HH EH1 N R IY0 ',\n",
- " 'T AH0 D EY1 F AY1 V Y IH1 R Z L EY1 T ER0 W IY1 AA1 R F EY1 S IH0 NG AH0 S IH1 M AH0 L ER0 S IH2 CH UW0 EY1 SH AH0 N ',\n",
- " 'W EH1 N AY1 S AO1 Y UW1 K IH1 S IH0 NG Y UW1 L UH1 K T R IH1 L IY0 HH AE1 P IY0 ',\n",
- " 'OW1 N L IY0 W AH1 N V IY1 HH IH0 K AH0 L M EY1 B IY1 AH0 L AW1 D T UW1 P AA1 R K AE1 T EH1 N IY0 G IH1 V AH0 N T AY1 M ',\n",
- " 'DH AH0 D EH1 D L AY2 N Z AA1 R IH2 N D IY1 D V EH1 R IY0 T AY1 T ',\n",
- " 'AY1 EH1 M G L AE1 D Y UW1 EH2 N JH OY1 D Y ER0 S EH1 L F ',\n",
- " 'W AH1 T AA1 R Y UW1 S T IH1 L D UW1 IH0 NG HH IY1 R ',\n",
- " 'DH IH1 S IH1 Z AE1 N AE1 N AH0 M AH0 L DH AE1 T IH1 Z AH0 D M AY1 ER0 D F AO1 R IH1 T S W AY1 T N AH0 S AH0 N D K L EH1 N L IY0 N IH0 S ',\n",
- " 'P ER0 HH AE1 P S DH EH1 R IH1 Z AH0 N AH1 DH ER0 W EY1 T UW1 P OW1 Z DH IY1 Z IH1 SH UW0 Z ',\n",
- " 'Y AO1 R S T UW1 D AH0 N T S T EH1 S T S K AO1 R Z D R AA1 P L OW1 ER0 AH0 N D L OW1 ER0 EH1 V ER0 IY0 Y IH1 R ',\n",
- " 'W EH0 R EH1 V ER0 HH ER1 T IH1 R Z F EH1 L AH0 F R UW1 T T R IY1 G R UW1 ',\n",
- " 'AY1 W AA1 Z AH0 B AW1 T T UW1 HH EH1 D B AE1 K T UW1 M AY1 HH OW0 T EH1 L AH0 N D G OW1 T UW1 S L IY1 P ',\n",
- " 'Y UW1 S EH1 D SH IY1 R IH1 L IY0 HH EH1 L P T L AE1 S T T AY1 M ',\n",
- " 'M AY1 F EY1 V ER0 IH0 T S IY1 Z AH0 N S P R IH1 NG IH1 Z HH IY1 R ',\n",
- " 'HH IY1 EH1 S DH AH0 R IH1 CH G AY1 HH UW1 B IH1 L T DH IY0 EH1 R P L EY0 N Z ',\n",
- " 'AA1 T OW2 AH0 N D IH0 L IH1 Z AH0 B AH0 TH G EY1 V IH1 T T UW1 AH1 S F AO1 R DH AH0 W EH1 D IH0 NG IH2 N K R EH1 D AH0 B L IY0 JH EH1 N ER0 AH0 S ',\n",
- " 'L UH1 K DH AH0 P AH0 L IY1 S S EH1 D DH AE1 T DH EH1 R W AA1 Z N AH1 TH IH0 NG S T OW1 L AH0 N F R AH1 M DH AH0 HH AW1 S ',\n",
- " 'AH0 N D AY1 S AH0 P OW1 Z W IY1 K AE1 N TH AE1 NG K Y AO1 R B R AH1 DH ER0 F AO1 R DH AE1 T ',\n",
- " 'DH AE1 T EH1 S EY0 P R IH1 T IY0 D EY1 N JH ER0 AH0 S TH IH1 NG Y UW1 R EY1 D UW1 IH0 NG ',\n",
- " 'HH IY1 ER0 AY1 V D IH0 N JH AH0 P AE1 N F AO1 R DH AH0 F ER1 S T T AY1 M AE1 T DH IY0 EY1 JH AH1 V T W EH1 N T IY0 S IH1 K S ',\n",
- " 'S AE1 M TH AO1 T W IY1 W ER1 HH AE1 V IH0 NG F AH1 N B IY1 IH0 NG T AH0 G EH1 DH ER0 ',\n",
- " 'W EH1 L DH AH0 T R UW1 V AE1 L Y UW0 AH1 V S AH1 M TH IH0 NG IH1 S N T IY1 AO1 L W EY2 Z D IH0 T ER1 M AH0 N D B AY1 IH1 T S P R AY1 S ',\n",
- " 'N OW1 IH1 T EH1 S N AA1 T P AH0 L AY1 T T UW1 D IH0 S K AH1 S AH0 L EY1 D IY0 EH1 S EY1 JH ',\n",
- " 'JH AH1 S T AH0 N AH1 DH ER0 K W AO1 R T ER0 M AY1 L AH0 N D AY1 D AA1 N T IY1 HH AE1 V T UW1 B IY1 T AA1 L ER0 AH0 N T EH1 V ER0 AH0 G EH1 N ',\n",
- " 'B AH1 T JH OW1 N Z AH0 P AA1 R T M AH0 N T HH AE1 D OW1 N L IY0 B IH1 N R EH1 N T IH0 D AW1 T F AO1 R AH0 W IY1 K ',\n",
- " 'W AH1 T Y AO1 R P ER1 F IH1 K T D EY1 W UH1 D HH AE1 V B IH1 N L AY1 K ',\n",
- " 'N AA1 T AH0 V EH1 R IY0 Y UW1 S F AH0 L S K IH1 L AH0 S P EH1 SH L IY0 W EH1 N DH AH0 M AH1 N IY0 R AH1 N Z AW1 T ',\n",
+ " ' D UW1 Y UW1 R IY1 AH0 L AY2 Z W AH1 T T AY1 M IH1 T IH1 Z ',\n",
+ " ' HH IY1 K AH1 M Z B AE1 K T UW1 DH AH0 V AE1 L IY0 ',\n",
+ " ' DH IH1 S D R EH1 S D AH1 Z N AA1 T L UH1 K W ER1 TH M AH1 CH ',\n",
+ " ' W AH1 T HH AE1 P AH0 N D T AH0 N AY1 T HH AE1 Z N AH1 TH IH0 NG T UW1 D UW1 W IH1 DH HH EH1 N R IY0 ',\n",
+ " ' T AH0 D EY1 F AY1 V Y IH1 R Z L EY1 T ER0 W IY1 AA1 R F EY1 S IH0 NG AH0 S IH1 M AH0 L ER0 S IH2 CH UW0 EY1 SH AH0 N ',\n",
+ " ' W EH1 N AY1 S AO1 Y UW1 K IH1 S IH0 NG Y UW1 L UH1 K T R IH1 L IY0 HH AE1 P IY0 ',\n",
+ " ' OW1 N L IY0 W AH1 N V IY1 HH IH0 K AH0 L M EY1 B IY1 AH0 L AW1 D T UW1 P AA1 R K AE1 T EH1 N IY0 G IH1 V AH0 N T AY1 M ',\n",
+ " ' DH AH0 D EH1 D L AY2 N Z AA1 R IH2 N D IY1 D V EH1 R IY0 T AY1 T ',\n",
+ " ' AY1 EH1 M G L AE1 D Y UW1 EH2 N JH OY1 D Y ER0 S EH1 L F ',\n",
+ " ' W AH1 T AA1 R Y UW1 S T IH1 L D UW1 IH0 NG HH IY1 R ',\n",
+ " ' DH IH1 S IH1 Z AE1 N AE1 N AH0 M AH0 L DH AE1 T IH1 Z AH0 D M AY1 ER0 D F AO1 R IH1 T S W AY1 T N AH0 S AH0 N D K L EH1 N L IY0 N IH0 S ',\n",
+ " ' P ER0 HH AE1 P S DH EH1 R IH1 Z AH0 N AH1 DH ER0 W EY1 T UW1 P OW1 Z DH IY1 Z IH1 SH UW0 Z ',\n",
+ " ' Y AO1 R S T UW1 D AH0 N T S T EH1 S T S K AO1 R Z D R AA1 P L OW1 ER0 AH0 N D L OW1 ER0 EH1 V ER0 IY0 Y IH1 R ',\n",
+ " ' W EH0 R EH1 V ER0 HH ER1 T IH1 R Z F EH1 L AH0 F R UW1 T T R IY1 G R UW1 ',\n",
+ " ' AY1 W AA1 Z AH0 B AW1 T T UW1 HH EH1 D B AE1 K T UW1 M AY1 HH OW0 T EH1 L AH0 N D G OW1 T UW1 S L IY1 P ',\n",
+ " ' Y UW1 S EH1 D SH IY1 R IH1 L IY0 HH EH1 L P T L AE1 S T T AY1 M ',\n",
+ " ' M AY1 F EY1 V ER0 IH0 T S IY1 Z AH0 N S P R IH1 NG IH1 Z HH IY1 R ',\n",
+ " ' HH IY1 EH1 S DH AH0 R IH1 CH G AY1 HH UW1 B IH1 L T DH IY0 EH1 R P L EY0 N Z ',\n",
+ " ' AA1 T OW2 AH0 N D IH0 L IH1 Z AH0 B AH0 TH G EY1 V IH1 T T UW1 AH1 S F AO1 R DH AH0 W EH1 D IH0 NG IH2 N K R EH1 D AH0 B L IY0 JH EH1 N ER0 AH0 S ',\n",
+ " ' L UH1 K DH AH0 P AH0 L IY1 S S EH1 D DH AE1 T DH EH1 R W AA1 Z N AH1 TH IH0 NG S T OW1 L AH0 N F R AH1 M DH AH0 HH AW1 S ',\n",
+ " ' AH0 N D AY1 S AH0 P OW1 Z W IY1 K AE1 N TH AE1 NG K Y AO1 R B R AH1 DH ER0 F AO1 R DH AE1 T ',\n",
+ " ' DH AE1 T EH1 S EY0 P R IH1 T IY0 D EY1 N JH ER0 AH0 S TH IH1 NG Y UW1 R EY1 D UW1 IH0 NG ',\n",
+ " ' HH IY1 ER0 AY1 V D IH0 N JH AH0 P AE1 N F AO1 R DH AH0 F ER1 S T T AY1 M AE1 T DH IY0 EY1 JH AH1 V T W EH1 N T IY0 S IH1 K S ',\n",
+ " ' S AE1 M TH AO1 T W IY1 W ER1 HH AE1 V IH0 NG F AH1 N B IY1 IH0 NG T AH0 G EH1 DH ER0 ',\n",
+ " ' W EH1 L DH AH0 T R UW1 V AE1 L Y UW0 AH1 V S AH1 M TH IH0 NG IH1 S N T IY1 AO1 L W EY2 Z D IH0 T ER1 M AH0 N D B AY1 IH1 T S P R AY1 S ',\n",
+ " ' N OW1 IH1 T EH1 S N AA1 T P AH0 L AY1 T T UW1 D IH0 S K AH1 S AH0 L EY1 D IY0 EH1 S EY1 JH ',\n",
+ " ' JH AH1 S T AH0 N AH1 DH ER0 K W AO1 R T ER0 M AY1 L AH0 N D AY1 D AA1 N T IY1 HH AE1 V T UW1 B IY1 T AA1 L ER0 AH0 N T EH1 V ER0 AH0 G EH1 N ',\n",
+ " ' B AH1 T JH OW1 N Z AH0 P AA1 R T M AH0 N T HH AE1 D OW1 N L IY0 B IH1 N R EH1 N T IH0 D AW1 T F AO1 R AH0 W IY1 K ',\n",
+ " ' W AH1 T Y AO1 R P ER1 F IH1 K T D EY1 W UH1 D HH AE1 V B IH1 N L AY1 K ',\n",
+ " ' N AA1 T AH0 V EH1 R IY0 Y UW1 S F AH0 L S K IH1 L AH0 S P EH1 SH L IY0 W EH1 N DH AH0 M AH1 N IY0 R AH1 N Z AW1 T ',\n",
"]"
]
},
@@ -1151,9 +1152,9 @@
" 1 | \n",
" Do you realize what time it is? | \n",
" 0011 | \n",
- " generated_hifi/esd_tune_reversal/g_2669999/0011/angry/1.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/1.wav | \n",
" generated_hifi/esd_tune/g_2669999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_2669999/0011/angry/1.wav | \n",
" Do you realize what time it is? | \n",
" \n",
" \n",
@@ -1169,9 +1170,9 @@
" | 3 | \n",
" What are you still doing here? | \n",
" 0011 | \n",
- " generated_hifi/esd_tune/g_2669999/0011/angry/10.wav | \n",
" generated_hifi/esd_tune_reversal/g_2669999/0011/angry/10.wav | \n",
" generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune/g_2669999/0011/angry/10.wav | \n",
" What are you still doing here? | \n",
"
\n",
" \n",
@@ -1196,9 +1197,9 @@
" | 2695 | \n",
" Only one vehicle may be allowed to park at any given time. | \n",
" 0020 | \n",
- " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/7.wav | \n",
- " generated_hifi/esd_tune/g_2669999/0020/surprise/7.wav | \n",
" generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune/g_2669999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/7.wav | \n",
" Only one vehicle may be allowed to park at any given time. | \n",
"
\n",
" \n",
@@ -1214,9 +1215,9 @@
" | 2697 | \n",
" The deadlines are indeed very tight. | \n",
" 0020 | \n",
- " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/8.wav | \n",
" generated_hifi/esd_tune/g_2669999/0020/surprise/8.wav | \n",
" generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/8.wav | \n",
" The deadlines are indeed very tight. | \n",
"
\n",
" \n",
@@ -1232,8 +1233,8 @@
" | 2699 | \n",
" I'm glad you enjoyed yourself. | \n",
" 0020 | \n",
- " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/9.wav | \n",
" generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/9.wav | \n",
" generated_hifi/esd_tune/g_2669999/0020/surprise/9.wav | \n",
" I'm glad you enjoyed yourself. | \n",
"
\n",
@@ -1271,40 +1272,40 @@
"\n",
" m1 \\\n",
"0 id_1 \n",
- "1 generated_hifi/esd_tune_reversal/g_2669999/0011/angry/1.wav \n",
+ "1 generated_hifi/esd_tune/g_2669999/0011/angry/1.wav \n",
"2 id_4 \n",
- "3 generated_hifi/esd_tune/g_2669999/0011/angry/10.wav \n",
+ "3 generated_hifi/esd_tune_reversal/g_2669999/0011/angry/10.wav \n",
"4 id_7 \n",
"... ... \n",
- "2695 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/7.wav \n",
+ "2695 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/7.wav \n",
"2696 id_4045 \n",
- "2697 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/8.wav \n",
+ "2697 generated_hifi/esd_tune/g_2669999/0020/surprise/8.wav \n",
"2698 id_4048 \n",
- "2699 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/9.wav \n",
+ "2699 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/9.wav \n",
"\n",
" m2 \\\n",
"0 id_2 \n",
"1 generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/1.wav \n",
"2 id_5 \n",
- "3 generated_hifi/esd_tune_reversal/g_2669999/0011/angry/10.wav \n",
+ "3 generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/10.wav \n",
"4 id_8 \n",
"... ... \n",
"2695 generated_hifi/esd_tune/g_2669999/0020/surprise/7.wav \n",
"2696 id_4046 \n",
- "2697 generated_hifi/esd_tune/g_2669999/0020/surprise/8.wav \n",
+ "2697 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/8.wav \n",
"2698 id_4049 \n",
- "2699 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/9.wav \n",
+ "2699 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/9.wav \n",
"\n",
" m3 \\\n",
"0 id_3 \n",
- "1 generated_hifi/esd_tune/g_2669999/0011/angry/1.wav \n",
+ "1 generated_hifi/esd_tune_reversal/g_2669999/0011/angry/1.wav \n",
"2 id_6 \n",
- "3 generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/10.wav \n",
+ "3 generated_hifi/esd_tune/g_2669999/0011/angry/10.wav \n",
"4 id_9 \n",
"... ... \n",
- "2695 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/7.wav \n",
+ "2695 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/7.wav \n",
"2696 id_4047 \n",
- "2697 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/8.wav \n",
+ "2697 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/8.wav \n",
"2698 id_4050 \n",
"2699 generated_hifi/esd_tune/g_2669999/0020/surprise/9.wav \n",
"\n",
@@ -1346,7 +1347,7 @@
},
{
"cell_type": "code",
- "execution_count": 15,
+ "execution_count": 14,
"id": "eef3b325-c26e-457a-b575-eb6b210aff7f",
"metadata": {},
"outputs": [],
@@ -1363,7 +1364,7 @@
},
{
"cell_type": "code",
- "execution_count": 16,
+ "execution_count": 15,
"id": "d32630c2-aca7-4ee9-b44c-a949d0fd17c4",
"metadata": {},
"outputs": [],
@@ -1380,7 +1381,7 @@
},
{
"cell_type": "code",
- "execution_count": 17,
+ "execution_count": 16,
"id": "2c11c3ca-abe4-4bbd-930f-8a4f62c9dc8f",
"metadata": {},
"outputs": [],
@@ -1403,7 +1404,7 @@
},
{
"cell_type": "code",
- "execution_count": 18,
+ "execution_count": 17,
"id": "11d84c76-841a-4215-a45f-268076e4ae21",
"metadata": {},
"outputs": [],
@@ -1415,7 +1416,7 @@
},
{
"cell_type": "code",
- "execution_count": 19,
+ "execution_count": 18,
"id": "26b577fe-3957-4902-8758-e5501f5d81a0",
"metadata": {},
"outputs": [],
@@ -1426,7 +1427,7 @@
},
{
"cell_type": "code",
- "execution_count": 20,
+ "execution_count": 19,
"id": "f37b7fb3-93eb-448c-a8a3-a23cb0d165ae",
"metadata": {},
"outputs": [],
@@ -1437,7 +1438,7 @@
},
{
"cell_type": "code",
- "execution_count": 21,
+ "execution_count": 20,
"id": "24aef771-215a-4fe6-82cb-1ff375681189",
"metadata": {
"scrolled": true,
@@ -1530,7 +1531,7 @@
},
{
"cell_type": "code",
- "execution_count": 22,
+ "execution_count": 21,
"id": "49a60285-31bf-4263-9162-c3d928aa9068",
"metadata": {},
"outputs": [
From 72b7bb953d81a2b2e43cac7c5c343bc75d9d67e0 Mon Sep 17 00:00:00 2001
From: ruk0sh <43880105+ruk0sh@users.noreply.github.com>
Date: Sat, 28 May 2022 10:52:53 +0300
Subject: [PATCH 17/17] Update HTML generation notebook
---
generate_listener_html.ipynb | 454 ++++++++++++++++++-----------------
1 file changed, 236 insertions(+), 218 deletions(-)
diff --git a/generate_listener_html.ipynb b/generate_listener_html.ipynb
index a23ed4d..3cf7791 100644
--- a/generate_listener_html.ipynb
+++ b/generate_listener_html.ipynb
@@ -17,7 +17,12 @@
"\n",
"\n",
"pd.set_option('display.max_colwidth', None)\n",
- "pd.set_option('display.colheader_justify', 'center')"
+ "pd.set_option('display.colheader_justify', 'center')\n",
+ "\n",
+ "\n",
+ "REPORTS_DIR = Path('reports')\n",
+ "REPORT_NAME = 'esd_tune_1-2-7'\n",
+ "OUTPUT_DIR = REPORTS_DIR / REPORT_NAME"
]
},
{
@@ -63,7 +68,7 @@
},
{
"cell_type": "code",
- "execution_count": 3,
+ "execution_count": null,
"id": "f0871808-7396-4a82-b529-404babb6be8d",
"metadata": {},
"outputs": [],
@@ -104,22 +109,22 @@
},
{
"cell_type": "code",
- "execution_count": 4,
+ "execution_count": 3,
"id": "d7942da8-a539-4a50-98ef-d93cbdcec403",
"metadata": {},
"outputs": [],
"source": [
"## NOTE: all model keys need to be composed of a single letter followed by a number\n",
"model2folder = {\n",
- " \"m1\": 'generated_hifi/esd_tune/g_2669999',\n",
- " \"m2\": 'generated_hifi/esd_tune_reversal/g_2669999',\n",
- " \"m3\": 'generated_hifi/esd_tune_advloss0/g_2669999',\n",
+ " \"m1\": 'generated_hifi/esd_tune_1/g_2879999',\n",
+ " \"m2\": 'generated_hifi/esd_tune_2/g_2939999',\n",
+ " \"m3\": 'generated_hifi/esd_tune_7/g_3059999',\n",
"}"
]
},
{
"cell_type": "code",
- "execution_count": 5,
+ "execution_count": 4,
"id": "d4d44f9e-525a-4e3d-bb5c-679b64bd7861",
"metadata": {},
"outputs": [],
@@ -129,7 +134,7 @@
},
{
"cell_type": "code",
- "execution_count": 6,
+ "execution_count": 5,
"id": "51e4eae6-ea64-4d69-947c-2a0c7b15678b",
"metadata": {},
"outputs": [
@@ -154,7 +159,7 @@
},
{
"cell_type": "code",
- "execution_count": 7,
+ "execution_count": 6,
"id": "e2031149-e1e8-421f-983c-83833c1b67ad",
"metadata": {},
"outputs": [],
@@ -166,7 +171,7 @@
},
{
"cell_type": "code",
- "execution_count": 8,
+ "execution_count": 7,
"id": "f60b3404-269a-407e-bedd-898b754c5651",
"metadata": {},
"outputs": [
@@ -419,7 +424,7 @@
"[2700 rows x 12 columns]"
]
},
- "execution_count": 8,
+ "execution_count": 7,
"metadata": {},
"output_type": "execute_result"
}
@@ -450,7 +455,7 @@
},
{
"cell_type": "code",
- "execution_count": 9,
+ "execution_count": 8,
"id": "2b88d46e-c3af-43eb-8af7-282492147a1f",
"metadata": {},
"outputs": [
@@ -507,9 +512,9 @@
" \n",
" \n",
" | 1 | \n",
- " generated_hifi/esd_tune/g_2669999/0011/angry/1.wav | \n",
- " generated_hifi/esd_tune_reversal/g_2669999/0011/angry/1.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_1/g_2879999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_2/g_2939999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_7/g_3059999/0011/angry/1.wav | \n",
" 1.wav | \n",
" 0011 | \n",
" Do you realize what time it is? | \n",
@@ -537,9 +542,9 @@
"
\n",
" \n",
" | 3 | \n",
- " generated_hifi/esd_tune/g_2669999/0011/angry/10.wav | \n",
- " generated_hifi/esd_tune_reversal/g_2669999/0011/angry/10.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune_1/g_2879999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune_2/g_2939999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune_7/g_3059999/0011/angry/10.wav | \n",
" 10.wav | \n",
" 0011 | \n",
" What are you still doing here? | \n",
@@ -582,9 +587,9 @@
"
\n",
" \n",
" | 2695 | \n",
- " generated_hifi/esd_tune/g_2669999/0020/surprise/7.wav | \n",
- " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/7.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune_1/g_2879999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune_2/g_2939999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune_7/g_3059999/0020/surprise/7.wav | \n",
" 7.wav | \n",
" 0020 | \n",
" Only one vehicle may be allowed to park at any given time. | \n",
@@ -612,9 +617,9 @@
"
\n",
" \n",
" | 2697 | \n",
- " generated_hifi/esd_tune/g_2669999/0020/surprise/8.wav | \n",
- " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/8.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_1/g_2879999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_2/g_2939999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_7/g_3059999/0020/surprise/8.wav | \n",
" 8.wav | \n",
" 0020 | \n",
" The deadlines are indeed very tight. | \n",
@@ -642,9 +647,9 @@
"
\n",
" \n",
" | 2699 | \n",
- " generated_hifi/esd_tune/g_2669999/0020/surprise/9.wav | \n",
- " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/9.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune_1/g_2879999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune_2/g_2939999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune_7/g_3059999/0020/surprise/9.wav | \n",
" 9.wav | \n",
" 0020 | \n",
" I'm glad you enjoyed yourself. | \n",
@@ -661,57 +666,57 @@
""
],
"text/plain": [
- " m1 \\\n",
- "0 id_1 \n",
- "1 generated_hifi/esd_tune/g_2669999/0011/angry/1.wav \n",
- "2 id_4 \n",
- "3 generated_hifi/esd_tune/g_2669999/0011/angry/10.wav \n",
- "4 id_7 \n",
- "... ... \n",
- "2695 generated_hifi/esd_tune/g_2669999/0020/surprise/7.wav \n",
- "2696 id_4045 \n",
- "2697 generated_hifi/esd_tune/g_2669999/0020/surprise/8.wav \n",
- "2698 id_4048 \n",
- "2699 generated_hifi/esd_tune/g_2669999/0020/surprise/9.wav \n",
+ " m1 \\\n",
+ "0 id_1 \n",
+ "1 generated_hifi/esd_tune_1/g_2879999/0011/angry/1.wav \n",
+ "2 id_4 \n",
+ "3 generated_hifi/esd_tune_1/g_2879999/0011/angry/10.wav \n",
+ "4 id_7 \n",
+ "... ... \n",
+ "2695 generated_hifi/esd_tune_1/g_2879999/0020/surprise/7.wav \n",
+ "2696 id_4045 \n",
+ "2697 generated_hifi/esd_tune_1/g_2879999/0020/surprise/8.wav \n",
+ "2698 id_4048 \n",
+ "2699 generated_hifi/esd_tune_1/g_2879999/0020/surprise/9.wav \n",
"\n",
- " m2 \\\n",
- "0 id_2 \n",
- "1 generated_hifi/esd_tune_reversal/g_2669999/0011/angry/1.wav \n",
- "2 id_5 \n",
- "3 generated_hifi/esd_tune_reversal/g_2669999/0011/angry/10.wav \n",
- "4 id_8 \n",
- "... ... \n",
- "2695 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/7.wav \n",
- "2696 id_4046 \n",
- "2697 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/8.wav \n",
- "2698 id_4049 \n",
- "2699 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/9.wav \n",
+ " m2 \\\n",
+ "0 id_2 \n",
+ "1 generated_hifi/esd_tune_2/g_2939999/0011/angry/1.wav \n",
+ "2 id_5 \n",
+ "3 generated_hifi/esd_tune_2/g_2939999/0011/angry/10.wav \n",
+ "4 id_8 \n",
+ "... ... \n",
+ "2695 generated_hifi/esd_tune_2/g_2939999/0020/surprise/7.wav \n",
+ "2696 id_4046 \n",
+ "2697 generated_hifi/esd_tune_2/g_2939999/0020/surprise/8.wav \n",
+ "2698 id_4049 \n",
+ "2699 generated_hifi/esd_tune_2/g_2939999/0020/surprise/9.wav \n",
"\n",
- " m3 \\\n",
- "0 id_3 \n",
- "1 generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/1.wav \n",
- "2 id_6 \n",
- "3 generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/10.wav \n",
- "4 id_9 \n",
- "... ... \n",
- "2695 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/7.wav \n",
- "2696 id_4047 \n",
- "2697 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/8.wav \n",
- "2698 id_4050 \n",
- "2699 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/9.wav \n",
+ " m3 m1_wavbasename \\\n",
+ "0 id_3 None \n",
+ "1 generated_hifi/esd_tune_7/g_3059999/0011/angry/1.wav 1.wav \n",
+ "2 id_6 None \n",
+ "3 generated_hifi/esd_tune_7/g_3059999/0011/angry/10.wav 10.wav \n",
+ "4 id_9 None \n",
+ "... ... ... \n",
+ "2695 generated_hifi/esd_tune_7/g_3059999/0020/surprise/7.wav 7.wav \n",
+ "2696 id_4047 None \n",
+ "2697 generated_hifi/esd_tune_7/g_3059999/0020/surprise/8.wav 8.wav \n",
+ "2698 id_4050 None \n",
+ "2699 generated_hifi/esd_tune_7/g_3059999/0020/surprise/9.wav 9.wav \n",
"\n",
- " m1_wavbasename m1_accent_speaker \\\n",
- "0 None None \n",
- "1 1.wav 0011 \n",
- "2 None None \n",
- "3 10.wav 0011 \n",
- "4 None None \n",
- "... ... ... \n",
- "2695 7.wav 0020 \n",
- "2696 None None \n",
- "2697 8.wav 0020 \n",
- "2698 None None \n",
- "2699 9.wav 0020 \n",
+ " m1_accent_speaker \\\n",
+ "0 None \n",
+ "1 0011 \n",
+ "2 None \n",
+ "3 0011 \n",
+ "4 None \n",
+ "... ... \n",
+ "2695 0020 \n",
+ "2696 None \n",
+ "2697 0020 \n",
+ "2698 None \n",
+ "2699 0020 \n",
"\n",
" m1_text \\\n",
"0 None \n",
@@ -781,7 +786,7 @@
"[2700 rows x 12 columns]"
]
},
- "execution_count": 9,
+ "execution_count": 8,
"metadata": {},
"output_type": "execute_result"
}
@@ -828,7 +833,7 @@
},
{
"cell_type": "code",
- "execution_count": 10,
+ "execution_count": 9,
"id": "efc3b273-7b7d-4165-aef8-38da88cfc154",
"metadata": {
"tags": []
@@ -858,7 +863,7 @@
},
{
"cell_type": "code",
- "execution_count": 11,
+ "execution_count": 10,
"id": "e3f303a2-0923-4b6a-802c-11c497fe5e58",
"metadata": {},
"outputs": [
@@ -868,7 +873,7 @@
"['m1', 'm2', 'm3']"
]
},
- "execution_count": 11,
+ "execution_count": 10,
"metadata": {},
"output_type": "execute_result"
}
@@ -880,7 +885,7 @@
},
{
"cell_type": "code",
- "execution_count": 12,
+ "execution_count": 11,
"id": "d1b37dd4-39c3-4956-adf0-22cf45b49f85",
"metadata": {},
"outputs": [
@@ -927,9 +932,9 @@
" 1 | \n",
" Do you realize what time it is? | \n",
" 0011 | \n",
- " generated_hifi/esd_tune/g_2669999/0011/angry/1.wav | \n",
- " generated_hifi/esd_tune_reversal/g_2669999/0011/angry/1.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_1/g_2879999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_2/g_2939999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_7/g_3059999/0011/angry/1.wav | \n",
" Do you realize what time it is? | \n",
"
\n",
" \n",
@@ -945,9 +950,9 @@
" | 3 | \n",
" What are you still doing here? | \n",
" 0011 | \n",
- " generated_hifi/esd_tune/g_2669999/0011/angry/10.wav | \n",
- " generated_hifi/esd_tune_reversal/g_2669999/0011/angry/10.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune_1/g_2879999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune_2/g_2939999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune_7/g_3059999/0011/angry/10.wav | \n",
" What are you still doing here? | \n",
"
\n",
" \n",
@@ -972,9 +977,9 @@
" | 2695 | \n",
" Only one vehicle may be allowed to park at any given time. | \n",
" 0020 | \n",
- " generated_hifi/esd_tune/g_2669999/0020/surprise/7.wav | \n",
- " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/7.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune_1/g_2879999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune_2/g_2939999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune_7/g_3059999/0020/surprise/7.wav | \n",
" Only one vehicle may be allowed to park at any given time. | \n",
"
\n",
" \n",
@@ -990,9 +995,9 @@
" | 2697 | \n",
" The deadlines are indeed very tight. | \n",
" 0020 | \n",
- " generated_hifi/esd_tune/g_2669999/0020/surprise/8.wav | \n",
- " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/8.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_1/g_2879999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_2/g_2939999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_7/g_3059999/0020/surprise/8.wav | \n",
" The deadlines are indeed very tight. | \n",
"
\n",
" \n",
@@ -1008,9 +1013,9 @@
" | 2699 | \n",
" I'm glad you enjoyed yourself. | \n",
" 0020 | \n",
- " generated_hifi/esd_tune/g_2669999/0020/surprise/9.wav | \n",
- " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/9.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune_1/g_2879999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune_2/g_2939999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune_7/g_3059999/0020/surprise/9.wav | \n",
" I'm glad you enjoyed yourself. | \n",
"
\n",
" \n",
@@ -1032,44 +1037,57 @@
"2698 None \n",
"2699 I'm glad you enjoyed yourself. \n",
"\n",
- " m1_accent_speaker m1 \\\n",
- "0 None id_1 \n",
- "1 0011 generated_hifi/esd_tune/g_2669999/0011/angry/1.wav \n",
- "2 None id_4 \n",
- "3 0011 generated_hifi/esd_tune/g_2669999/0011/angry/10.wav \n",
- "4 None id_7 \n",
- "... ... ... \n",
- "2695 0020 generated_hifi/esd_tune/g_2669999/0020/surprise/7.wav \n",
- "2696 None id_4045 \n",
- "2697 0020 generated_hifi/esd_tune/g_2669999/0020/surprise/8.wav \n",
- "2698 None id_4048 \n",
- "2699 0020 generated_hifi/esd_tune/g_2669999/0020/surprise/9.wav \n",
+ " m1_accent_speaker \\\n",
+ "0 None \n",
+ "1 0011 \n",
+ "2 None \n",
+ "3 0011 \n",
+ "4 None \n",
+ "... ... \n",
+ "2695 0020 \n",
+ "2696 None \n",
+ "2697 0020 \n",
+ "2698 None \n",
+ "2699 0020 \n",
+ "\n",
+ " m1 \\\n",
+ "0 id_1 \n",
+ "1 generated_hifi/esd_tune_1/g_2879999/0011/angry/1.wav \n",
+ "2 id_4 \n",
+ "3 generated_hifi/esd_tune_1/g_2879999/0011/angry/10.wav \n",
+ "4 id_7 \n",
+ "... ... \n",
+ "2695 generated_hifi/esd_tune_1/g_2879999/0020/surprise/7.wav \n",
+ "2696 id_4045 \n",
+ "2697 generated_hifi/esd_tune_1/g_2879999/0020/surprise/8.wav \n",
+ "2698 id_4048 \n",
+ "2699 generated_hifi/esd_tune_1/g_2879999/0020/surprise/9.wav \n",
"\n",
- " m2 \\\n",
- "0 id_2 \n",
- "1 generated_hifi/esd_tune_reversal/g_2669999/0011/angry/1.wav \n",
- "2 id_5 \n",
- "3 generated_hifi/esd_tune_reversal/g_2669999/0011/angry/10.wav \n",
- "4 id_8 \n",
- "... ... \n",
- "2695 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/7.wav \n",
- "2696 id_4046 \n",
- "2697 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/8.wav \n",
- "2698 id_4049 \n",
- "2699 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/9.wav \n",
+ " m2 \\\n",
+ "0 id_2 \n",
+ "1 generated_hifi/esd_tune_2/g_2939999/0011/angry/1.wav \n",
+ "2 id_5 \n",
+ "3 generated_hifi/esd_tune_2/g_2939999/0011/angry/10.wav \n",
+ "4 id_8 \n",
+ "... ... \n",
+ "2695 generated_hifi/esd_tune_2/g_2939999/0020/surprise/7.wav \n",
+ "2696 id_4046 \n",
+ "2697 generated_hifi/esd_tune_2/g_2939999/0020/surprise/8.wav \n",
+ "2698 id_4049 \n",
+ "2699 generated_hifi/esd_tune_2/g_2939999/0020/surprise/9.wav \n",
"\n",
- " m3 \\\n",
- "0 id_3 \n",
- "1 generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/1.wav \n",
- "2 id_6 \n",
- "3 generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/10.wav \n",
- "4 id_9 \n",
- "... ... \n",
- "2695 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/7.wav \n",
- "2696 id_4047 \n",
- "2697 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/8.wav \n",
- "2698 id_4050 \n",
- "2699 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/9.wav \n",
+ " m3 \\\n",
+ "0 id_3 \n",
+ "1 generated_hifi/esd_tune_7/g_3059999/0011/angry/1.wav \n",
+ "2 id_6 \n",
+ "3 generated_hifi/esd_tune_7/g_3059999/0011/angry/10.wav \n",
+ "4 id_9 \n",
+ "... ... \n",
+ "2695 generated_hifi/esd_tune_7/g_3059999/0020/surprise/7.wav \n",
+ "2696 id_4047 \n",
+ "2697 generated_hifi/esd_tune_7/g_3059999/0020/surprise/8.wav \n",
+ "2698 id_4050 \n",
+ "2699 generated_hifi/esd_tune_7/g_3059999/0020/surprise/9.wav \n",
"\n",
" m2_text \n",
"0 None \n",
@@ -1087,7 +1105,7 @@
"[2700 rows x 6 columns]"
]
},
- "execution_count": 12,
+ "execution_count": 11,
"metadata": {},
"output_type": "execute_result"
}
@@ -1105,7 +1123,7 @@
},
{
"cell_type": "code",
- "execution_count": 13,
+ "execution_count": 12,
"id": "a9915d2e-01e5-40c0-923b-4bf6bf7389cf",
"metadata": {},
"outputs": [
@@ -1152,9 +1170,9 @@
" 1 | \n",
" Do you realize what time it is? | \n",
" 0011 | \n",
- " generated_hifi/esd_tune/g_2669999/0011/angry/1.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/1.wav | \n",
- " generated_hifi/esd_tune_reversal/g_2669999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_2/g_2939999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_7/g_3059999/0011/angry/1.wav | \n",
+ " generated_hifi/esd_tune_1/g_2879999/0011/angry/1.wav | \n",
" Do you realize what time it is? | \n",
" \n",
" \n",
@@ -1170,9 +1188,9 @@
" | 3 | \n",
" What are you still doing here? | \n",
" 0011 | \n",
- " generated_hifi/esd_tune_reversal/g_2669999/0011/angry/10.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/10.wav | \n",
- " generated_hifi/esd_tune/g_2669999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune_7/g_3059999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune_2/g_2939999/0011/angry/10.wav | \n",
+ " generated_hifi/esd_tune_1/g_2879999/0011/angry/10.wav | \n",
" What are you still doing here? | \n",
"
\n",
" \n",
@@ -1197,9 +1215,9 @@
" | 2695 | \n",
" Only one vehicle may be allowed to park at any given time. | \n",
" 0020 | \n",
- " generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/7.wav | \n",
- " generated_hifi/esd_tune/g_2669999/0020/surprise/7.wav | \n",
- " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune_2/g_2939999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune_7/g_3059999/0020/surprise/7.wav | \n",
+ " generated_hifi/esd_tune_1/g_2879999/0020/surprise/7.wav | \n",
" Only one vehicle may be allowed to park at any given time. | \n",
"
\n",
" \n",
@@ -1215,9 +1233,9 @@
" | 2697 | \n",
" The deadlines are indeed very tight. | \n",
" 0020 | \n",
- " generated_hifi/esd_tune/g_2669999/0020/surprise/8.wav | \n",
- " generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/8.wav | \n",
- " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_7/g_3059999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_1/g_2879999/0020/surprise/8.wav | \n",
+ " generated_hifi/esd_tune_2/g_2939999/0020/surprise/8.wav | \n",
" The deadlines are indeed very tight. | \n",
"
\n",
" \n",
@@ -1233,9 +1251,9 @@
" | 2699 | \n",
" I'm glad you enjoyed yourself. | \n",
" 0020 | \n",
- " generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/9.wav | \n",
- " generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/9.wav | \n",
- " generated_hifi/esd_tune/g_2669999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune_2/g_2939999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune_1/g_2879999/0020/surprise/9.wav | \n",
+ " generated_hifi/esd_tune_7/g_3059999/0020/surprise/9.wav | \n",
" I'm glad you enjoyed yourself. | \n",
"
\n",
" \n",
@@ -1270,44 +1288,44 @@
"2698 - \n",
"2699 0020 \n",
"\n",
- " m1 \\\n",
- "0 id_1 \n",
- "1 generated_hifi/esd_tune/g_2669999/0011/angry/1.wav \n",
- "2 id_4 \n",
- "3 generated_hifi/esd_tune_reversal/g_2669999/0011/angry/10.wav \n",
- "4 id_7 \n",
- "... ... \n",
- "2695 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/7.wav \n",
- "2696 id_4045 \n",
- "2697 generated_hifi/esd_tune/g_2669999/0020/surprise/8.wav \n",
- "2698 id_4048 \n",
- "2699 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/9.wav \n",
+ " m1 \\\n",
+ "0 id_1 \n",
+ "1 generated_hifi/esd_tune_2/g_2939999/0011/angry/1.wav \n",
+ "2 id_4 \n",
+ "3 generated_hifi/esd_tune_7/g_3059999/0011/angry/10.wav \n",
+ "4 id_7 \n",
+ "... ... \n",
+ "2695 generated_hifi/esd_tune_2/g_2939999/0020/surprise/7.wav \n",
+ "2696 id_4045 \n",
+ "2697 generated_hifi/esd_tune_7/g_3059999/0020/surprise/8.wav \n",
+ "2698 id_4048 \n",
+ "2699 generated_hifi/esd_tune_2/g_2939999/0020/surprise/9.wav \n",
"\n",
- " m2 \\\n",
- "0 id_2 \n",
- "1 generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/1.wav \n",
- "2 id_5 \n",
- "3 generated_hifi/esd_tune_advloss0/g_2669999/0011/angry/10.wav \n",
- "4 id_8 \n",
- "... ... \n",
- "2695 generated_hifi/esd_tune/g_2669999/0020/surprise/7.wav \n",
- "2696 id_4046 \n",
- "2697 generated_hifi/esd_tune_advloss0/g_2669999/0020/surprise/8.wav \n",
- "2698 id_4049 \n",
- "2699 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/9.wav \n",
+ " m2 \\\n",
+ "0 id_2 \n",
+ "1 generated_hifi/esd_tune_7/g_3059999/0011/angry/1.wav \n",
+ "2 id_5 \n",
+ "3 generated_hifi/esd_tune_2/g_2939999/0011/angry/10.wav \n",
+ "4 id_8 \n",
+ "... ... \n",
+ "2695 generated_hifi/esd_tune_7/g_3059999/0020/surprise/7.wav \n",
+ "2696 id_4046 \n",
+ "2697 generated_hifi/esd_tune_1/g_2879999/0020/surprise/8.wav \n",
+ "2698 id_4049 \n",
+ "2699 generated_hifi/esd_tune_1/g_2879999/0020/surprise/9.wav \n",
"\n",
- " m3 \\\n",
- "0 id_3 \n",
- "1 generated_hifi/esd_tune_reversal/g_2669999/0011/angry/1.wav \n",
- "2 id_6 \n",
- "3 generated_hifi/esd_tune/g_2669999/0011/angry/10.wav \n",
- "4 id_9 \n",
- "... ... \n",
- "2695 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/7.wav \n",
- "2696 id_4047 \n",
- "2697 generated_hifi/esd_tune_reversal/g_2669999/0020/surprise/8.wav \n",
- "2698 id_4050 \n",
- "2699 generated_hifi/esd_tune/g_2669999/0020/surprise/9.wav \n",
+ " m3 \\\n",
+ "0 id_3 \n",
+ "1 generated_hifi/esd_tune_1/g_2879999/0011/angry/1.wav \n",
+ "2 id_6 \n",
+ "3 generated_hifi/esd_tune_1/g_2879999/0011/angry/10.wav \n",
+ "4 id_9 \n",
+ "... ... \n",
+ "2695 generated_hifi/esd_tune_1/g_2879999/0020/surprise/7.wav \n",
+ "2696 id_4047 \n",
+ "2697 generated_hifi/esd_tune_2/g_2939999/0020/surprise/8.wav \n",
+ "2698 id_4050 \n",
+ "2699 generated_hifi/esd_tune_7/g_3059999/0020/surprise/9.wav \n",
"\n",
" m2_text \n",
"0 - \n",
@@ -1325,7 +1343,7 @@
"[2700 rows x 6 columns]"
]
},
- "execution_count": 13,
+ "execution_count": 12,
"metadata": {},
"output_type": "execute_result"
}
@@ -1352,9 +1370,9 @@
"metadata": {},
"outputs": [],
"source": [
- "Path('reports').mkdir(exist_ok=True)\n",
+ "OUTPUT_DIR.mkdir(exist_ok=True, parents=True)\n",
"df_permuted.to_csv(\n",
- " 'reports/eval_map.csv',\n",
+ " OUTPUT_DIR / 'eval_map.csv',\n",
" sep = \"\\t\",\n",
" header = False,\n",
" index = False,\n",
@@ -1432,7 +1450,7 @@
"metadata": {},
"outputs": [],
"source": [
- "output_folder = Path('reports/html')\n",
+ "output_folder = OUTPUT_DIR / 'html'\n",
"output_folder.mkdir(parents=True, exist_ok=True)"
]
},
@@ -1450,59 +1468,59 @@
"output_type": "stream",
"text": [
"0 100\n",
- "reports/html/nat_vctk_esd_tune_test_1.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_1.html\n",
"100 200\n",
- "reports/html/nat_vctk_esd_tune_test_2.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_2.html\n",
"200 300\n",
- "reports/html/nat_vctk_esd_tune_test_3.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_3.html\n",
"300 400\n",
- "reports/html/nat_vctk_esd_tune_test_4.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_4.html\n",
"400 500\n",
- "reports/html/nat_vctk_esd_tune_test_5.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_5.html\n",
"500 600\n",
- "reports/html/nat_vctk_esd_tune_test_6.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_6.html\n",
"600 700\n",
- "reports/html/nat_vctk_esd_tune_test_7.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_7.html\n",
"700 800\n",
- "reports/html/nat_vctk_esd_tune_test_8.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_8.html\n",
"800 900\n",
- "reports/html/nat_vctk_esd_tune_test_9.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_9.html\n",
"900 1000\n",
- "reports/html/nat_vctk_esd_tune_test_10.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_10.html\n",
"1000 1100\n",
- "reports/html/nat_vctk_esd_tune_test_11.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_11.html\n",
"1100 1200\n",
- "reports/html/nat_vctk_esd_tune_test_12.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_12.html\n",
"1200 1300\n",
- "reports/html/nat_vctk_esd_tune_test_13.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_13.html\n",
"1300 1400\n",
- "reports/html/nat_vctk_esd_tune_test_14.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_14.html\n",
"1400 1500\n",
- "reports/html/nat_vctk_esd_tune_test_15.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_15.html\n",
"1500 1600\n",
- "reports/html/nat_vctk_esd_tune_test_16.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_16.html\n",
"1600 1700\n",
- "reports/html/nat_vctk_esd_tune_test_17.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_17.html\n",
"1700 1800\n",
- "reports/html/nat_vctk_esd_tune_test_18.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_18.html\n",
"1800 1900\n",
- "reports/html/nat_vctk_esd_tune_test_19.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_19.html\n",
"1900 2000\n",
- "reports/html/nat_vctk_esd_tune_test_20.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_20.html\n",
"2000 2100\n",
- "reports/html/nat_vctk_esd_tune_test_21.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_21.html\n",
"2100 2200\n",
- "reports/html/nat_vctk_esd_tune_test_22.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_22.html\n",
"2200 2300\n",
- "reports/html/nat_vctk_esd_tune_test_23.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_23.html\n",
"2300 2400\n",
- "reports/html/nat_vctk_esd_tune_test_24.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_24.html\n",
"2400 2500\n",
- "reports/html/nat_vctk_esd_tune_test_25.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_25.html\n",
"2500 2600\n",
- "reports/html/nat_vctk_esd_tune_test_26.html\n",
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_26.html\n",
"2600 2700\n",
- "reports/html/nat_vctk_esd_tune_test_27.html\n"
+ "reports/esd_tune_1-2-7/html/nat_vctk_esd_tune_test_27.html\n"
]
}
],
@@ -1539,8 +1557,8 @@
"name": "stdout",
"output_type": "stream",
"text": [
- "reports/scores/scores.csv\n",
- "reports/scores/scores.xlsx\n"
+ "reports/esd_tune_1-2-7/scores/scores.csv\n",
+ "reports/esd_tune_1-2-7/scores/scores.xlsx\n"
]
}
],
@@ -1569,7 +1587,7 @@
" print(excel_path)\n",
"\n",
"\n",
- "create_scores_file(df_permuted, \"reports/scores\")"
+ "create_scores_file(df_permuted, OUTPUT_DIR / \"scores\")"
]
}
],