Voice-first session loop for DeepSeek Harness
PerryLink/dsh-talk · productivity
Voice-first session loop for DeepSeek Harness: a composer microphone button with browser/local speech-to-text (Web Speech, FunASR, whisper.cpp), a speak tool for text-to-speech replies (browser, edge-tts, piper), event a.
★ 9GitHub stars
2Forks
2026-09-08Last updated
TypeScriptLanguage
Apache-2.0License
Key features
- Voice-first session loop for DeepSeek Harness: talk to it, hear it answer.
- speak tool — the agent speaks its replies aloud.
- Permissionsthe plugin stores nothing but an in-memory, byte-capped audio cache
- Model-visible ⟺ logged — the model sees only the speak tool's canonical value and render text.
- Press the mic, speak, and the reply is spoken back — with speak-to-interrupt.
Requirements
- Browser supportWeb Speech and MediaRecorder are feature-detected
- without them the mic button disables itself and host engines (FunASR/whisper.cpp) can still transcribe when configured.
Install command
dsh plugin --profile web add github:PerryLink/dsh-talk