Xinference use chattts BUG

[NieSf]

System Info / 系統信息

curl -X 'POST'
'http://192.168.100.251:9997/v1/audio/speech'
-H 'accept: application/json'
-H 'Content-Type: application/json'
-d '{
"model": "ChatTTS",
"input": "你好",
"voice": "Echo",
"response_format":"wav"
}'

By using WAV format, speech can be generated very successfully. But if the request format is changed to mp3, only the following garbled code will be obtained:
ID3#TSSELavf60.16.100ÿ俉nfoLavf

Running Xinference with Docker? / 是否使用 Docker 运行 Xinfernece？

• docker / docker
• pip install / 通过 pip install 安装
• installation from source / 从源码安装

Version info / 版本信息

xinference, version 0.16.2

The command used to start Xinference / 用以启动 xinference 的命令

xinference-local --host 192.168.100.251 --port 9997 --log-level debug

Reproduction / 复现过程

curl -X 'POST'
'http://192.168.100.251:9997/v1/audio/speech'
-H 'accept: application/json'
-H 'Content-Type: application/json'
-d '{
"model": "ChatTTS",
"input": "你好",
"voice": "Echo",
"response_format":"mp3"
}'

Expected behavior / 期待表现

This problem has been bothering me for a long time, and I hope some friends can give me some help

[NieSf]

When requesting MP3 format, the correct response cannot be obtained. In this case, xinference debug mode does not output any errors