@try-works/dsh-stt
Multi-model local speech-to-text for the DSH Web composer: Whistle and Voz, user-selectable, with lazy per-model download. Ships as a DSH bundle. 适合需要在本地进行语音输入，并支持按需延迟下载模型的使用者。
설치
$
npx -p @deepseek-ai/dsh dsh plugin --profile web add github:try-works/dsh-sttConfiguration
cordis.patch.yml carries the whole configuration. Each model is a key under providers:
providers:
whistle-local:
enabled: true
voz-local:
enabled: false # keep 390 MB out of the picker entirely
sensevoice-local:
enabled: true # placeholder only
| Field | Meaning |
|---|---|
enabled | false stops the provider registering — it is never listed, selected or downloaded |
modelRoot | where that model's files live; defaults to / |
id | override the provider id (advanced; the selection is stored by id) |
name | override the display name shown in the picker |
dataRoot is required and must be an absolute path.