CroissanTTs/dsh-voice-mini ↗★ 0
dsh-voice-mini
Voice-feedback plugin for DeepSeek Harness: a speak tool, readReplies narration, an LLM verbalizer pass, per-session voices, chimes, a monitoring panel, and a native macOS floating pet. zh/en i18n. 适合希望听语音反馈或朗读回复的用户；含 macOS 悬浮宠物。
설치
$
npx -p @deepseek-ai/dsh dsh plugin --profile web add github:CroissanTTs/dsh-voice-miniConfiguration
Key fields (all overridable from the settings modal or POST /voice-mini/config):
| Field | Default | Purpose |
|---|---|---|
preset | default | instant / default / quiet |
backend | edge | edge / kokoro / say / fake |
voiceMode | per-session | per-session auto-assign or fixed |
voicePalette | [] | custom voice pool; empty = engine default (8 zh 4F4M + 2 en) |
readReplies | false | narrate messages verbatim |
chimeEnabled / chimeSpeech / chimeStatus | true / glass / soft | earcon system |
statusSpeech | true | quiet preset: chime-only (false) vs. spoken (true) |
phraseTurnEnd | "" | custom completion phrase; {title} = workspace dir name; empty = locale default |
locale | zh | zh / en — switches UI, spoken phrases, verbalizer prompt |
summarizeResult / summarizeProvider / summarizeModel | true / "" / "" | verbalizer pass (empty provider = auto-detect first adapter; empty model = stays on the template phrase until you pick one in the panel) |