API requests results are not split text and returns broken speech (Koboldcpp)

[Battou2501]

So I am trying to use OpenAI API mode in Koboldcpp to use Chatterbox server. It sends requests and I see chatterbox generating responses. BUT! There sre two issues.

1. It does nut split text to chunks and always process text in one go. This may be causing second issue
2. Returned audio contains broken speech. Like the person talkin have a stroke and speak random words with ocasional words from text.

Have someone encountered this? Is there any fix?

[devnen]

Fixed in b9862f0. The /v1/audio/speech endpoint now automatically splits long text into chunks and stitches them with crossfade, same as the /tts endpoint. This should resolve the garbled speech on longer inputs.