Text to Speech (TTS)

Enable TTS

ask --tts "summarize this"

Modes

Full mode

Speaks after full model response is complete:

ask --tts --tts-configure full "explain this"

Stream mode

Speaks while tokens are being generated:

ask --tts --tts-configure stream "read this while generating"

Runtime Requirements

• piper executable on PATH
• aplay executable on PATH
• valid Piper model file

Default model path:

• ~/.local/share/piper/en_US-lessac-medium.onnx

Relevant Environment Variables

• ASK_TTS_MODEL
• ASK_TTS_SAMPLE_RATE
• ASK_TTS_MODE
• ASK_TTS_DELAY_SECONDS (buffer delay, clamped to ~100-300ms)

Implementation Notes

Current streaming behavior includes:

• small delay buffer for smoother cadence
• tiny chunk merging to reduce choppiness
• flush on Piper stdin after each chunk
• Ctrl+C interrupt support for immediate stop
• lock-based protection against overlapping audio playback