From 5e94b9652df4b6dad711beaffd4e061d3c098c5d Mon Sep 17 00:00:00 2001
From: sean1572 <shperry@ucsd.edu>
Date: Tue, 13 Jan 2026 15:48:53 -0800
Subject: [PATCH 1/2] Inital commit of label studio exporter

---
 .gitignore                                    |   2 +-
 .../label_studio_exporter/README.md           |  24 ++
 data_exporters/label_studio_exporter/demo.py  |  47 ++++
 .../label_studio_exporter/label_studio.py     | 231 ++++++++++++++++++
 .../label_studio_exporter/template.xml        |  12 +
 5 files changed, 315 insertions(+), 1 deletion(-)
 create mode 100644 data_exporters/label_studio_exporter/README.md
 create mode 100644 data_exporters/label_studio_exporter/demo.py
 create mode 100644 data_exporters/label_studio_exporter/label_studio.py
 create mode 100644 data_exporters/label_studio_exporter/template.xml

diff --git a/.gitignore b/.gitignore
index 30db9cd..a31c918 100644
--- a/.gitignore
+++ b/.gitignore
@@ -237,4 +237,4 @@ demos/
 # Block predictions
 predictions/*
 *.pkl
-*.arrow
\ No newline at end of file
+*.arrow
diff --git a/data_exporters/label_studio_exporter/README.md b/data_exporters/label_studio_exporter/README.md
new file mode 100644
index 0000000..088436a
--- /dev/null
+++ b/data_exporters/label_studio_exporter/README.md
@@ -0,0 +1,24 @@
+# Labeling Audio Data in Label Studio
+
+This pipeline intends to take audiodatasets from the inferance or the data_extractors of `whoot_model_training` and format projects in label studio to easily verify annotations. The following readme outlines best practices for creating and intergrating with Label Studio
+
+
+## Creating new project
+
+1) Create a .env file with the following properties in the same folder that this readme is in
+
+```
+# Define the URL where Label Studio is accessible
+LABEL_STUDIO_URL = 'HOSTNAME OF LABEL STUDIO INSTANCE'
+# API key is available at the Account & Settings page in Label Studio UI
+LABEL_STUDIO_API_KEY = 'INSERT YOUR API KEY'
+LABEL_STUDIO_PROJECT_ID = 'PROJECT ID FROM URL OF PROJECT'
+```
+
+2) Create a new project in that label studio instance and uploaded the needed data to it
+
+NOTE: Save the project_id from the URL of the project
+
+If keeping the data local on the instance, try to keep the file structure the same as is the audio file from your ML machine. For example, if some dataset is located at `mnt/datasets/audio_dataset_cool/AB/1/audio.wav` then you may want to make the path on label studio something like `label_studio_path/audio_dataset_cool/AB/1/audio.wav` for the easiest intergrations. Otherwise some minor file changes will be needed.
+
+3) Run the script to apply annotations, see demo.py in this folder
\ No newline at end of file
diff --git a/data_exporters/label_studio_exporter/demo.py b/data_exporters/label_studio_exporter/demo.py
new file mode 100644
index 0000000..b524fc9
--- /dev/null
+++ b/data_exporters/label_studio_exporter/demo.py
@@ -0,0 +1,47 @@
+"""Demo for using Label Studio Exporter with a sample dataset."""
+
+from label_studio import LabelStudioSetup
+from dotenv import load_dotenv
+import os   
+
+load_dotenv()
+
+
+## SELECT A PROJECT FROM LABEL STUDIO
+## FIND ID IN URL OF PROJECT
+PROJECT_ID = int(os.getenv("LABEL_STUDIO_PROJECT_ID"))
+ls_setup = LabelStudioSetup(
+    current_project=PROJECT_ID
+)
+## ADD DEFAULT TEMPLATE TO LABEL STUDIO
+ls_setup.apply_custom_template("template.xml")
+
+## HOW TO GET AUDIO FILES TO REVIEW
+# Note this is not a perfect process as diffrences between label studio and your dataset may exist
+ls_setup.get_files(ls_file_parent='data/local-files/?d=data1/')
+
+## TODO MAKE SURE YOUR DATASET AND LABEL STUDIO FILE PATHS ALIGN
+
+#===============================================================
+## below is a fake dataset creation for demo purposes only
+## In practice you would load your dataset from the saves in whoot_model_training
+import datasets
+import random
+
+class_list = ['cluck','coocoo', 'twitter', 'alarm', 'chick begging', 'no_buow']
+
+ds = datasets.Dataset.from_dict({
+    "audio": ls_setup.get_files(ls_file_parent='data/local-files/?d=data1/')["files"],
+    "labels": random.choices(class_list, k=len(ls_setup.get_files(ls_file_parent='data/local-files/?d=data1/')["files"]))
+})
+
+ds = ds.cast_column("audio", datasets.Audio(sampling_rate=16000, decode=False))
+
+#===============================================================
+
+
+
+## UPLOAD DATASET TO LABEL STUDIO
+ls_setup.update_tasks_in_ls(ds, ls_file_parent='data/local-files/?d=data1/', is_model_prediction=True)
+
+
diff --git a/data_exporters/label_studio_exporter/label_studio.py b/data_exporters/label_studio_exporter/label_studio.py
new file mode 100644
index 0000000..cf53016
--- /dev/null
+++ b/data_exporters/label_studio_exporter/label_studio.py
@@ -0,0 +1,231 @@
+from http import client
+import os
+import requests
+from dotenv import load_dotenv
+from label_studio_sdk import LabelStudio
+import tqdm
+from label_studio_sdk.label_interface.objects import PredictionValue, AnnotationValue
+import datasets
+
+class LabelStudioSetup():
+    """Sets up a Label Studio project for annotation.
+    
+    When submoduling, primarly do so for diffrent labeling templates. In particular,
+    - apply_audio_template
+    - default_template_annotation_style
+
+    These will be template spefific. Currently, they mirror the template found in
+    data_exporters/label_studio_exporter/template.xml
+    """
+
+    def __init__(self, current_project = None):
+        """Initialize the Label Studio client and create a project."""
+        load_dotenv()
+        LABEL_STUDIO_URL = os.getenv("LABEL_STUDIO_URL")
+        LABEL_STUDIO_API_KEY = os.getenv("LABEL_STUDIO_API_KEY")
+
+        
+        if LABEL_STUDIO_URL is None or LABEL_STUDIO_API_KEY is None:
+            raise ValueError("LABEL_STUDIO_URL and LABEL_STUDIO_API_KEY must be set in the .env file.")
+
+        self.client = LabelStudio(base_url=LABEL_STUDIO_URL, api_key=LABEL_STUDIO_API_KEY)
+
+        if current_project is not None:
+            self.current_project = self.client.projects.get(current_project)
+            print("Project ID:", self.current_project.id, "\t Project Name:", self.current_project.title)
+            input("Double check this, this script can take destructive actions. Press Enter to continue...")
+        
+        self.api_key = LABEL_STUDIO_API_KEY
+        self.base_url = LABEL_STUDIO_URL
+
+    def create_project(self, title: str = "Whoot Audio Annotation Project"):
+        """Create a new project in Label Studio.
+
+        Args:
+            title (str): The title of the project.
+
+        Returns:
+            project: The created Label Studio project.
+        """
+        project = self.client.projects.create(
+            title=title,
+            label_config=self.label_config
+        )
+
+        print("Project ID:", project.id, project.title)
+        input("Double check this, this script can take destructive actions. Press Enter to continue...")
+        
+        # Associate this class instance with the created project
+        self.current_project = project
+
+        return project
+
+    def apply_audio_template(self, class_names: list = None):
+        """Apply a default audio annotation template.
+        
+        Args:
+            class_names (list): List of class names for labeling.
+        """
+        
+        if self.current_project is None:
+            raise ValueError("No current project set. Please create a project first.")
+        
+        audio_template = """
+        <View>
+        <Labels name="labels" toName="audio">
+        """
+
+        for class_name in class_names:
+            audio_template += f'<Label value="{class_name}"/>\n'
+
+        audio_template += """
+        </Labels>
+        <Audio name="audio" value="$audio" decoder="ffmpeg" spectrogram="true" height="500"/>
+        """      
+
+        print("Applying audio annotation template to project ID:", self.current_project.id)
+        print(audio_template)  
+
+        self.set_label_interface(template=audio_template)
+        self.xml_template = audio_template
+
+    def apply_custom_template(self, template_path: str):
+        """Apply a custom annotation template from a file.
+
+        Args:
+            template_path (str): Path to the XML template file.
+        """
+        if self.current_project is None:
+            raise ValueError("No current project set. Please create a project first.")
+
+        with open(template_path, "r") as f:
+            custom_template = f.read()
+
+        print("Applying custom annotation template to project ID:", self.current_project.id)
+        print(custom_template)
+        self.set_label_interface(template=custom_template)
+        self.xml_template = custom_template
+
+    def set_label_interface(self, template: str):
+        req = self.client.projects.update(
+            id=self.current_project.id,
+            label_config=template
+        )
+        print(req)
+
+    def get_files(self, ls_file_parent: str = None):
+        """Retrieve all files in the current project.
+
+        Returns:
+            list: List of files in the project.
+        """
+        if self.current_project is None:
+            raise ValueError("No current project set. Please create a project first.")
+
+        files = []
+        ids = []
+        response = self.client.tasks.list(project=self.current_project.id)
+        for item in response:
+            audio_path = item.data["audio"]
+            audio_path = audio_path.replace(self.base_url, "")
+            if ls_file_parent is not None:
+                audio_path = audio_path.replace(ls_file_parent, "")
+            files.append(audio_path)
+            ids.append(item.id)
+        return {"files": files, "ids": ids}
+
+    def add_labels_to_task(self, task_id: int, result: list, prediction=True):
+        """Update all tasks in the current project.
+
+        Args:
+            tasks (list): List of task dictionaries to update.
+        """
+        if self.current_project is None:
+            raise ValueError("No current project set. Please create a project first.")
+
+        if prediction:
+            self.client.predictions.update(
+                id=self.current_project.id,
+                task=task_id,
+                result=result,
+            )
+        else:
+            self.client.annotations.update(
+                id=self.current_project.id,
+                task=task_id,
+                ground_truth=True,
+                result=result,
+            )
+
+
+    def default_template_annotation_style(
+            self,
+            id: int,
+            offset: float,
+            duration: float,
+            label: str,
+            file_path: str,
+            prediction: bool = False
+        ):
+
+        label_type = "annotations"
+        if prediction:
+            label_type = "predictions"
+
+        return {
+            "id": id,
+            f"{label_type}":
+                {
+                    "from_name": "labels",
+                    "to_name": "audio",
+                    "type": "labels",
+                    "value": {
+                        "start": offset,
+                        "end": offset + duration,
+                        "labels": [label]
+                    }
+                }
+            ,
+            "data": {
+                "audio": file_path
+            }
+        }
+
+    def update_tasks_in_ls(self, ds: datasets.Dataset, ls_file_parent: str, is_model_prediction=True):
+        li = self.current_project.get_label_interface()
+        files = self.get_files(ls_file_parent=ls_file_parent)["files"]
+        task_ids = self.get_files(ls_file_parent=ls_file_parent)["ids"]
+
+        datasets.disable_progress_bars()
+
+        for i in tqdm.tqdm(range(len(files)), desc="Updating tasks in project: {}".format(self.current_project.title)):
+            id = task_ids[i]
+            file_ds = ds.filter(lambda x: x['audio']['path'] == files[i])
+
+            # Our custom configuration of datasets allow for segmentation labels in audio :)
+            # This checks for it
+            if "offset" in file_ds[0]["audio"]:
+                offset = file_ds[0]["audio"]["offset"]
+                duration = file_ds[0]["audio"]["duration"]
+            else:
+                offset = 0.0
+                duration = 1.0
+
+
+            file_ds = file_ds.map(lambda x: self.default_template_annotation_style(
+                id, offset, duration, x['labels'], x['audio']['path'], prediction=is_model_prediction),)
+
+            if is_model_prediction:
+                prediction = PredictionValue(
+                    # # Tag predictions with specific model version
+                    model_version='my_model_v1',
+                    result=file_ds["predictions"]
+                )
+                self.client.predictions.create(task=id, **prediction.model_dump())
+            else:
+                annotations = AnnotationValue(
+                    # Define your labels here
+                    result=file_ds["annotations"]
+                )
+                self.client.annotations.create(task=id, **annotations.model_dump())
+        datasets.enable_progress_bars()
\ No newline at end of file
diff --git a/data_exporters/label_studio_exporter/template.xml b/data_exporters/label_studio_exporter/template.xml
new file mode 100644
index 0000000..22a4fe1
--- /dev/null
+++ b/data_exporters/label_studio_exporter/template.xml
@@ -0,0 +1,12 @@
+<View>
+  <Labels name="labels" toName="audio">
+    <Label value="cluck"/>
+    <Label value="coocoo"/>
+    <Label value="twitter"/>
+    <Label value="alarm"/>
+    <Label value="chick begging"/>
+    <Label value="no_buow"/>
+  </Labels>
+
+  <Audio name="audio" value="$audio" decoder="ffmpeg" spectrogram="true" height="500"/>
+</View>
\ No newline at end of file

From b6a188051ff4c3880c1e4b728a6be254f8d697cf Mon Sep 17 00:00:00 2001
From: sean1572 <shperry@ucsd.edu>
Date: Fri, 16 Jan 2026 16:54:58 -0800
Subject: [PATCH 2/2] Linted

---
 data_exporters/label_studio_exporter/demo.py  |  93 ++++----
 .../label_studio_exporter/label_studio.py     | 216 ++++++++++++------
 2 files changed, 201 insertions(+), 108 deletions(-)

diff --git a/data_exporters/label_studio_exporter/demo.py b/data_exporters/label_studio_exporter/demo.py
index b524fc9..d8b2ed1 100644
--- a/data_exporters/label_studio_exporter/demo.py
+++ b/data_exporters/label_studio_exporter/demo.py
@@ -1,47 +1,56 @@
 """Demo for using Label Studio Exporter with a sample dataset."""
 
-from label_studio import LabelStudioSetup
+import os
+import random
 from dotenv import load_dotenv
-import os   
-
-load_dotenv()
-
-
-## SELECT A PROJECT FROM LABEL STUDIO
-## FIND ID IN URL OF PROJECT
-PROJECT_ID = int(os.getenv("LABEL_STUDIO_PROJECT_ID"))
-ls_setup = LabelStudioSetup(
-    current_project=PROJECT_ID
-)
-## ADD DEFAULT TEMPLATE TO LABEL STUDIO
-ls_setup.apply_custom_template("template.xml")
-
-## HOW TO GET AUDIO FILES TO REVIEW
-# Note this is not a perfect process as diffrences between label studio and your dataset may exist
-ls_setup.get_files(ls_file_parent='data/local-files/?d=data1/')
-
-## TODO MAKE SURE YOUR DATASET AND LABEL STUDIO FILE PATHS ALIGN
-
-#===============================================================
-## below is a fake dataset creation for demo purposes only
-## In practice you would load your dataset from the saves in whoot_model_training
+from label_studio import LabelStudioSetup
 import datasets
-import random
-
-class_list = ['cluck','coocoo', 'twitter', 'alarm', 'chick begging', 'no_buow']
-
-ds = datasets.Dataset.from_dict({
-    "audio": ls_setup.get_files(ls_file_parent='data/local-files/?d=data1/')["files"],
-    "labels": random.choices(class_list, k=len(ls_setup.get_files(ls_file_parent='data/local-files/?d=data1/')["files"]))
-})
-
-ds = ds.cast_column("audio", datasets.Audio(sampling_rate=16000, decode=False))
-
-#===============================================================
-
-
-
-## UPLOAD DATASET TO LABEL STUDIO
-ls_setup.update_tasks_in_ls(ds, ls_file_parent='data/local-files/?d=data1/', is_model_prediction=True)
-
 
+if __name__ == "__main__":
+
+    load_dotenv()
+
+    # SELECT A PROJECT FROM LABEL STUDIO
+    # FIND ID IN URL OF PROJECT
+    PROJECT_ID = int(os.getenv("LABEL_STUDIO_PROJECT_ID"))
+    ls_setup = LabelStudioSetup(
+        current_project=PROJECT_ID
+    )
+    # ADD DEFAULT TEMPLATE TO LABEL STUDIO
+    ls_setup.apply_custom_template("template.xml")
+
+    # HOW TO GET AUDIO FILES TO REVIEW
+    # Note this is not a perfect process as
+    # diffrences between label studio and your dataset may exist
+    ls_setup.get_files(ls_file_parent='data/local-files/?d=data1/')
+
+    # Make sure your file names align to label studio files
+
+    # ===============================================================
+    # below is a fake dataset creation for demo purposes only
+    # In practice you would load your dataset from the saves in
+    # whoot_model_training
+    class_list = ['cluck', 'coocoo',
+                  'twitter', 'alarm', 'chick begging', 'no_buow']
+
+    ds = datasets.Dataset.from_dict({
+        "audio": ls_setup.get_files(
+            ls_file_parent='data/local-files/?d=data1/')["files"],
+        "labels": random.choices(
+            class_list, k=len(
+                ls_setup.get_files(
+                    ls_file_parent='data/local-files/?d=data1/')["files"]
+            )
+        )
+    })
+
+    ds = ds.cast_column(
+        "audio", datasets.Audio(sampling_rate=16000, decode=False))
+    # ===============================================================
+
+    # UPLOAD DATASET TO LABEL STUDIO
+    ls_setup.update_tasks_in_ls(
+        ds,
+        ls_file_parent='data/local-files/?d=data1/',
+        is_model_prediction=True
+    )
diff --git a/data_exporters/label_studio_exporter/label_studio.py b/data_exporters/label_studio_exporter/label_studio.py
index cf53016..9a15cab 100644
--- a/data_exporters/label_studio_exporter/label_studio.py
+++ b/data_exporters/label_studio_exporter/label_studio.py
@@ -1,42 +1,62 @@
-from http import client
+"""Wrapper for handling Label Studio API."""
 import os
-import requests
 from dotenv import load_dotenv
 from label_studio_sdk import LabelStudio
-import tqdm
-from label_studio_sdk.label_interface.objects import PredictionValue, AnnotationValue
+from label_studio_sdk.label_interface.objects import (
+    PredictionValue, AnnotationValue
+)
 import datasets
+import tqdm
+
 
 class LabelStudioSetup():
     """Sets up a Label Studio project for annotation.
-    
-    When submoduling, primarly do so for diffrent labeling templates. In particular,
+
+    When submoduling, primarly do so for diffrent labeling templates.
+    In particular,
     - apply_audio_template
     - default_template_annotation_style
 
-    These will be template spefific. Currently, they mirror the template found in
+    These will be template spefific.
+    Currently, they mirror the template found in
     data_exporters/label_studio_exporter/template.xml
     """
 
-    def __init__(self, current_project = None):
+    def __init__(self, current_project=None):
         """Initialize the Label Studio client and create a project."""
         load_dotenv()
-        LABEL_STUDIO_URL = os.getenv("LABEL_STUDIO_URL")
-        LABEL_STUDIO_API_KEY = os.getenv("LABEL_STUDIO_API_KEY")
+        label_studio_url = os.getenv("LABEL_STUDIO_URL")
+        label_studio_api_key = os.getenv("LABEL_STUDIO_API_KEY")
 
-        
-        if LABEL_STUDIO_URL is None or LABEL_STUDIO_API_KEY is None:
-            raise ValueError("LABEL_STUDIO_URL and LABEL_STUDIO_API_KEY must be set in the .env file.")
+        if label_studio_url is None or label_studio_api_key is None:
+            raise ValueError(
+                "LABEL_STUDIO_URL and LABEL_STUDIO_API_KEY",
+                "must be set in the .env file."
+            )
 
-        self.client = LabelStudio(base_url=LABEL_STUDIO_URL, api_key=LABEL_STUDIO_API_KEY)
+        self.client = LabelStudio(
+            base_url=label_studio_url,
+            api_key=label_studio_api_key
+        )
 
         if current_project is not None:
             self.current_project = self.client.projects.get(current_project)
-            print("Project ID:", self.current_project.id, "\t Project Name:", self.current_project.title)
-            input("Double check this, this script can take destructive actions. Press Enter to continue...")
-        
-        self.api_key = LABEL_STUDIO_API_KEY
-        self.base_url = LABEL_STUDIO_URL
+            print(
+                "Project ID:",
+                self.current_project.id,
+                "\t Project Name:",
+                self.current_project.title
+            )
+            input(
+                "Double check this,",
+                "this script can take destructive actions.",
+                "Press Enter to continue..."
+            )
+
+        self.api_key = label_studio_api_key
+        self.base_url = label_studio_url
+        self.xml_template = None
+        self.label_config = None
 
     def create_project(self, title: str = "Whoot Audio Annotation Project"):
         """Create a new project in Label Studio.
@@ -53,38 +73,49 @@ def create_project(self, title: str = "Whoot Audio Annotation Project"):
         )
 
         print("Project ID:", project.id, project.title)
-        input("Double check this, this script can take destructive actions. Press Enter to continue...")
-        
+        input(
+            "Double check this,",
+            "this script can take destructive actions.",
+            "Press Enter to continue..."
+        )
+
         # Associate this class instance with the created project
         self.current_project = project
-
         return project
 
     def apply_audio_template(self, class_names: list = None):
         """Apply a default audio annotation template.
-        
+
         Args:
             class_names (list): List of class names for labeling.
         """
-        
         if self.current_project is None:
-            raise ValueError("No current project set. Please create a project first.")
-        
+            raise ValueError(
+                "No current project set. Please create a project first."
+            )
+
         audio_template = """
         <View>
         <Labels name="labels" toName="audio">
-        """
-
-        for class_name in class_names:
-            audio_template += f'<Label value="{class_name}"/>\n'
+            <Label value="""
+        # pylint: disable=consider-using-join
+        audio_template += '\"/>\n\t <Label value="'.join(
+            class_names
+        ) + '\"/>\n'
 
         audio_template += """
         </Labels>
-        <Audio name="audio" value="$audio" decoder="ffmpeg" spectrogram="true" height="500"/>
-        """      
-
-        print("Applying audio annotation template to project ID:", self.current_project.id)
-        print(audio_template)  
+        <Audio
+            name="audio"
+            value="$audio"
+            decoder="ffmpeg"
+            spectrogram="true"
+            height="500"
+        />"""
+
+        print("Applying audio annotation template to project ID:",
+              self.current_project.id)
+        print(audio_template)
 
         self.set_label_interface(template=audio_template)
         self.xml_template = audio_template
@@ -96,17 +127,27 @@ def apply_custom_template(self, template_path: str):
             template_path (str): Path to the XML template file.
         """
         if self.current_project is None:
-            raise ValueError("No current project set. Please create a project first.")
+            raise ValueError(
+                "No current project set. Please create a project first."
+            )
 
-        with open(template_path, "r") as f:
+        with open(template_path, "r", encoding="utf-8") as f:
             custom_template = f.read()
 
-        print("Applying custom annotation template to project ID:", self.current_project.id)
+        print("Applying custom annotation template to project ID:",
+              self.current_project.id)
         print(custom_template)
         self.set_label_interface(template=custom_template)
         self.xml_template = custom_template
 
     def set_label_interface(self, template: str):
+        """Updates labeling interface for current project.
+
+        Default template at template.xml
+
+        Args:
+            template (str): XML string defining the labeling interface.
+        """
         req = self.client.projects.update(
             id=self.current_project.id,
             label_config=template
@@ -120,7 +161,9 @@ def get_files(self, ls_file_parent: str = None):
             list: List of files in the project.
         """
         if self.current_project is None:
-            raise ValueError("No current project set. Please create a project first.")
+            raise ValueError(
+                "No current project set. Please create a project first."
+            )
 
         files = []
         ids = []
@@ -138,10 +181,15 @@ def add_labels_to_task(self, task_id: int, result: list, prediction=True):
         """Update all tasks in the current project.
 
         Args:
-            tasks (list): List of task dictionaries to update.
+            task_id (int): Task ID in label studio.
+            result (list): List of label dictionaries to add.
+            prediction (bool): Whether this is a prediction or annotation.
+
         """
         if self.current_project is None:
-            raise ValueError("No current project set. Please create a project first.")
+            raise ValueError(
+                "No current project set. Please create a project first."
+            )
 
         if prediction:
             self.client.predictions.update(
@@ -157,25 +205,35 @@ def add_labels_to_task(self, task_id: int, result: list, prediction=True):
                 result=result,
             )
 
-
     def default_template_annotation_style(
-            self,
-            id: int,
-            offset: float,
-            duration: float,
-            label: str,
-            file_path: str,
-            prediction: bool = False
-        ):
+        self,
+        task_id: int,
+        offset: float,
+        duration: float,
+        label: str,
+        file_path: str,
+        prediction: bool = False
+    ):
+        """Applies a default annotation template for any audio dataset.
+
+        Args:
+            id (int): Task ID in label studio per audio file
+            offset (float): Start time of the label.
+            duration (float): Duration of the label.
+            label (str): Label value.
+            file_path (str): Path to the audio file.
+            prediction (bool): Whether this is a prediction or annotation.
 
+        Returns:
+            dict: Dictionary formatted for Label Studio API.
+        """
         label_type = "annotations"
         if prediction:
             label_type = "predictions"
 
         return {
-            "id": id,
-            f"{label_type}":
-                {
+            "id": task_id,
+            f"{label_type}": {
                     "from_name": "labels",
                     "to_name": "audio",
                     "type": "labels",
@@ -184,25 +242,42 @@ def default_template_annotation_style(
                         "end": offset + duration,
                         "labels": [label]
                     }
-                }
-            ,
+            },
             "data": {
                 "audio": file_path
             }
         }
 
-    def update_tasks_in_ls(self, ds: datasets.Dataset, ls_file_parent: str, is_model_prediction=True):
-        li = self.current_project.get_label_interface()
+    def update_tasks_in_ls(
+            self,
+            ds: datasets.Dataset,
+            ls_file_parent: str,
+            is_model_prediction=True):
+        """Update tasks in label studio with data from a dataset.
+
+        Args:
+            ds (datasets.Dataset): AudioDataset containing audio and labels.
+                See whoot_model_training/whoot_model_training/dataset/
+                audio_dataset.py
+            ls_file_parent (str): Parent path in label studio for audio files.
+            is_model_prediction (bool): Whether the labels are model
+                predictions.
+        """
+        # li = self.current_project.get_label_interface()
         files = self.get_files(ls_file_parent=ls_file_parent)["files"]
         task_ids = self.get_files(ls_file_parent=ls_file_parent)["ids"]
 
         datasets.disable_progress_bars()
 
-        for i in tqdm.tqdm(range(len(files)), desc="Updating tasks in project: {}".format(self.current_project.title)):
-            id = task_ids[i]
+        for i in tqdm.tqdm(
+            range(len(files)),
+            desc=f"Updating tasks in project: {self.current_project.title}"
+        ):
+            task_id = task_ids[i]
             file_ds = ds.filter(lambda x: x['audio']['path'] == files[i])
 
-            # Our custom configuration of datasets allow for segmentation labels in audio :)
+            # Our custom configuration of datasets
+            # allow for segmentation labels in audio :)
             # This checks for it
             if "offset" in file_ds[0]["audio"]:
                 offset = file_ds[0]["audio"]["offset"]
@@ -211,9 +286,16 @@ def update_tasks_in_ls(self, ds: datasets.Dataset, ls_file_parent: str, is_model
                 offset = 0.0
                 duration = 1.0
 
-
-            file_ds = file_ds.map(lambda x: self.default_template_annotation_style(
-                id, offset, duration, x['labels'], x['audio']['path'], prediction=is_model_prediction),)
+            file_ds = file_ds.map(
+                lambda x: self.default_template_annotation_style(
+                    task_id,
+                    offset,
+                    duration,
+                    x['labels'],
+                    x['audio']['path'],
+                    prediction=is_model_prediction
+                )
+            )
 
             if is_model_prediction:
                 prediction = PredictionValue(
@@ -221,11 +303,13 @@ def update_tasks_in_ls(self, ds: datasets.Dataset, ls_file_parent: str, is_model
                     model_version='my_model_v1',
                     result=file_ds["predictions"]
                 )
-                self.client.predictions.create(task=id, **prediction.model_dump())
+                self.client.predictions.create(
+                    task=task_id, **prediction.model_dump())
             else:
                 annotations = AnnotationValue(
                     # Define your labels here
                     result=file_ds["annotations"]
                 )
-                self.client.annotations.create(task=id, **annotations.model_dump())
-        datasets.enable_progress_bars()
\ No newline at end of file
+                self.client.annotations.create(
+                    task=task_id, **annotations.model_dump())
+        datasets.enable_progress_bars()