Skip to content

Instantly share code, notes, and snippets.

@rubiojr
Created January 16, 2026 13:34
Show Gist options
  • Select an option

  • Save rubiojr/7e761f697d2560de012822669eef9976 to your computer and use it in GitHub Desktop.

Select an option

Save rubiojr/7e761f697d2560de012822669eef9976 to your computer and use it in GitHub Desktop.
TTS Skill
name description
tts
Use pocket-tts to speak responses aloud via text-to-speech.

Prerequisites

A pocket-tts server must be running:

uvx pocket-tts serve

How to Speak

Use this command to generate and play audio:

curl -sX POST http://localhost:8000/tts -F "text=Your message here" -o /tmp/out.wav && \
  PULSE_SERVER=/run/user/$(id -u)/pulse/native mpv --no-terminal --ao=pulse /tmp/out.wav

Usage Guidelines

  • Use TTS when the user asks you to speak or read responses aloud
  • Keep messages concise for better audio experience
  • The server uses the default voice unless a custom voice is specified

Custom Voice

To use a different voice, add the voice_url parameter:

curl -sX POST http://localhost:8000/tts \
  -F "text=Hello" \
  -F "voice_url=hf://kyutai/tts-voices/marius/casual.wav" \
  -o /tmp/out.wav && \
  PULSE_SERVER=/run/user/$(id -u)/pulse/native mpv --no-terminal --ao=pulse /tmp/out.wav

Available voices: alba, marius, javert, jean, fantine, cosette, eponine, azelma

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment