Skip to content

Audio

KoboldCPP supports both Text to Speech and Speech to Text.

Audio Settings

  • Whisper Model

    • OpenAI when they still were "Open" under the early leadership of Musk and Altman before Musk left released a model to convert Audio to text in several different sizes.
    • https://huggingface.co/koboldcpp/whisper contains usable versions of the smaller models in a single location
    • This expects a GGML .bin file rather than a GGUF file.

  • OuteTTS Model (Text-to-Speech)

    • Converts audio tokens to speech, can sometimes work with singing and similar as well, but not nearly as good as a model focused on music
    • https://huggingface.co/koboldcpp/tts contains prequanted OuteTTS 1b and 500m

  • OuteTTS Threads

  • OuteTTS Max Tokens

    • How many tokens should be generated and convert to speech

  • WavTokenizer Model (Text-To-Speech)

    • Converts text to audio tokens usable by OuteTTS

  • TTS Use GPU

    • Use the GPU for this process instead of CPU. Useful for faster audio responses, but takes up space that could be used for a larger text model