[BUG] Type is not msgpack serializable: TTSRequest

[talongao]

I am encountering issues while attempting to use the SDK for Text-to-Speech (TTS). Despite following the provided documentation and examples, I am unable to successfully generate speech from text.

Steps to Reproduce:

1. Setup Environment:

   • Operating System: Ubuntu 20.04
   • SDK Version: 2024.10.22
   • Programming Language: Python 3.10.15

2. Code Snippet:

from fish_audio_sdk import Session, TTSRequest

def main():
    f = None
    if result_path is not None:
        f = open(result_path, "wb")
    for chunk in self.session.tts(
        TTSRequest(text=text, reference_id=voice_id, format=audio_format)
    ):
        if f is not None:
            f.write(chunk)
            f.flush()
        elif result_array is not None:
            result_array.extend(chunk)

if __name__ == "__main__":
    main()

3. Expected Behavior:

• The SDK should successfully synthesize the provided text into speech and return an audio stream.

4. Actual Behavior:

• raise Exception: TypeError('Type is not msgpack serializable: TTSRequest')

[talongao]

It does not work in version 2024.11.28 either.

[abersheeran]

aviramha/ormsgpack#311

[talongao]

Thanks

[talongao]

The ormsgpack is still at version 1.6.0. If you are using pydantic >= 2.10.0, you can temporarily replace TTSRequest with a dict.It's work.

for chunk in self.session.tts(dict(text=text, reference_id=voice_id, format=audio_format)):
    f.write(chunk)
    f.flush()