feat: support live TTS of fish audio

[frostming]

Signed-off-by: Frost Ming [email protected]

[yihong0618]

voice always the same
@frostming no matter how to change the voice id

[frostming]

It seems like an upstream issue. The voice is not changeable in websocket mode via reference_id config @abersheeran

[abersheeran]

It seems like an upstream issue. The voice is not changeable in websocket mode via reference_id config @abersheeran

已经复现，转交给相关同事在看了。

[abersheeran]

验证过了，已经修复。

[yihong0618]

现在的情况是不播放了。。。

[abersheeran]

测了几遍都没问题，能不能看看具体是什么错误？网络？返回的二进制不是音频流？还是提前中断？

[frostming]

测了几遍都没问题，能不能看看具体是什么错误？网络？返回的二进制不是音频流？还是提前中断？

Traceback (most recent call last):
  File "/Users/fming/wkspace/github/tetos/.venv/lib/python3.11/site-packages/fish_audio_sdk/websocket.py", line 135, in tts
    message = await ws.receive_bytes()
              ^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/fming/wkspace/github/tetos/.venv/lib/python3.11/site-packages/httpx_ws/_api.py", line 883, in receive_bytes
    event = await self.receive(timeout)
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/fming/wkspace/github/tetos/.venv/lib/python3.11/site-packages/httpx_ws/_api.py", line 805, in receive
    raise WebSocketDisconnect(event.code, event.reason)
httpx_ws._exceptions.WebSocketDisconnect: (<CloseReason.NO_STATUS_RCVD: 1005>, '')

[abersheeran]

明白了，你是手写的一个参数字典。之前 api 校验时默认值没处理好导致缺了个参数无法通过校验，现在滚动更新上去了。

[yihong0618]

@abersheeran @frostming Thanks