Only the first sentence is read when using the standard output as output file

[rkrisztian]

As the following examples show, when the text to read turn into voice contains multiple sentences, and I am using the standard output as output file, piper would fail:

(venv) myuser@my-linux:~/piper$ echo "First sentence." | piper --model en_US-amy-medium.onnx --output-file - | wc -c
58924

(venv) myuser@my-linux:~/piper$ echo "First sentence. Second sentence." | piper --model en_US-amy-medium.onnx --output-file - | wc -c
Traceback (most recent call last):
  File "/home/myuser/piper/venv/lib/python3.12/site-packages/piper/__main__.py", line 151, in main
    voice.synthesize(text, wav_file, **synthesize_args)
  File "/home/myuser/piper/venv/lib/python3.12/site-packages/piper/voice.py", line 103, in synthesize
    wav_file.writeframes(audio_bytes)
  File "/usr/lib/python3.12/wave.py", line 578, in writeframes
    self._patchheader()
  File "/usr/lib/python3.12/wave.py", line 633, in _patchheader
    curpos = self._file.tell()
             ^^^^^^^^^^^^^^^^^
OSError: [Errno 29] Illegal seek

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/myuser/piper/venv/bin/piper", line 8, in <module>
    sys.exit(main())
             ^^^^^^
  File "/home/myuser/piper/venv/lib/python3.12/site-packages/piper/__main__.py", line 150, in main
    with wave.open(sys.stdout.buffer, "wb") as wav_file:
  File "/usr/lib/python3.12/wave.py", line 471, in __exit__
    self.close()
  File "/usr/lib/python3.12/wave.py", line 585, in close
    self._patchheader()
  File "/usr/lib/python3.12/wave.py", line 633, in _patchheader
    curpos = self._file.tell()
             ^^^^^^^^^^^^^^^^^
OSError: [Errno 29] Illegal seek
121900

(venv) myuser@my-linux:~/piper$ echo "First sentence.\nSecond sentence." | piper --model en_US-amy-medium.onnx --output-file - | wc -c
148012

(venv) myuser@my-linux:~/piper$ echo "First sentence. Second sentence." | piper --model en_US-amy-medium.onnx --output-file my.wav && ls -lh my.wav && rm my.wav
-rw-rw-r-- 1 myuser myuser 119K Jun 28 23:50 my.wav

I really need this to work, because I'm trying to use Piper in Open WebUI through a Python script as OpenAI Speech API endpoint, and I can't accept writing to temporary files.

Thanks in advance.

PS.: Installing piper-tts was already a challenge, due to no compatible piper-phonemize version being available, see: rhasspy/piper-phonemize#33 (comment) .