BUG ChatTTS Internal Server Error

[halexan]

Describe the bug

ChatTTS部署成功，界面可以看到。

日志： 2024-07-04 00:42:25,572 xinference.model.utils 97 INFO Use model cache from a different hub. 2024-07-04 00:42:26,514 xinference.thirdparty.ChatTTS.core 16784 INFO Load from local: /root/.xinference/cache/ChatTTS 2024-07-04 00:42:27,649 xinference.thirdparty.ChatTTS.core 16784 INFO use cuda:0 2024-07-04 00:42:28,084 xinference.thirdparty.ChatTTS.core 16784 INFO vocos loaded. 2024-07-04 00:42:28,158 xinference.thirdparty.ChatTTS.core 16784 INFO dvae loaded. 2024-07-04 00:42:31,827 xinference.thirdparty.ChatTTS.core 16784 INFO gpt loaded. 2024-07-04 00:42:32,877 xinference.thirdparty.ChatTTS.core 16784 INFO decoder loaded. 2024-07-04 00:42:32,889 xinference.thirdparty.ChatTTS.core 16784 INFO tokenizer loaded. 2024-07-04 00:42:32,889 xinference.thirdparty.ChatTTS.core 16784 INFO All initialized.

但使用curl

curl -X 'POST' \
  'http://0.0.0.0:9997/v1/audio/speech' \
  -H 'accept: application/json' \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "ChatTTS",
    "text": "Hello",
    "voice": "echo",
  }'

返回： Internal Server Error

To Reproduce

To help us to reproduce this bug, please provide information below:

1. Your Python version. 3.10.13
2. The version of xinference you use. 0.12.3
3. Versions of crucial packages.
4. Full stack of the error.
5. Minimized code to reproduce the error.
6. docker image xprobe/xinference:v0.12.3

Expected behavior

返回语音mp3