dddzzz123-dz/dsh-voice-input-plugin

dddzzz123-dz★ 0JavaScriptLast synced: 2026-08-15

Open on GitHub

Voice input plugin for DeepSeek Harness Web UI.

README excerpt

DSH Voice Input Plugin 中文说明 A Cordis plugin for DeepSeek Harness (DSH) that adds continuous voice dictation and optional cloud ASR to the DSH Web UI composer. Features - Continuous voice dictation for the DSH composer. - Manual Chinese / English recognition switch ( zh-CN / en-US ). - Live interim transcript updates in the draft input. - User edits remain authoritative during live dictation; deleted or corrected text is not restored by a later final recognition event. - Auto-restart after browser-imposed silence endings. - Five-bar animated voice activity indicator. - DSH-style inline SVG microphone and stop icons, no emoji UI. - Optional high-accuracy cloud transcription through SiliconFlow. - All three Volcengine console ASR products are selectable: streaming, recording-file 1.0, and recording-file 2.0. - A package-private DSH Host proxy for Volcengine, avoiding browser CORS limitations and keeping credentials out of command lines and source files. Requirements - DeepSeek Harness / DSH with Cordis dynamic plugin support. - A DSH cordis / creator-mode session for installation. - Chrome or Edge. The plugin uses the browser Web Speech API: - SpeechRecognition / webkitSpeechRecogniti…

View full README on GitHub →
Skins / UIcodexdeepseek-harnessdsh-pluginvoice-inputweb-speechweb ui

Category