dsh-web-speech-input
leozou320-ai
Voice-to-text for the DeepSeek Harness Web UI — live, editable, never auto-sends. | DeepSeek Harness 网页语音输入
PROJECT TOPICS
PROJECT README

为 DeepSeek Harness Web 输入框添加中英语音输入,支持中文、英文及中英混合口述。基于浏览器 Web Speech API,无服务端、无需 API Key 或本地模型。
在 deepseek-harness 源码根目录执行:
pnpm dsh plugin --profile web add github:wepar1212/dsh-voice-input#v0.1.0
重启 DSH Web 后,点击输入框左侧的 🎤 开始持续识别;再次点击停止,并将识别结果写入输入框。
zh-CN;中文、英文和中英混合语音的实际准确率取决于浏览器的语音识别服务。0.1.0-rc.5 上验证。Adds Chinese and English voice input to the DeepSeek Harness Web composer, including mixed Chinese-English dictation. It uses the browser Web Speech API and requires no server, API key, or local speech model.
Run this command from the deepseek-harness source root:
pnpm dsh plugin --profile web add github:wepar1212/dsh-voice-input#v0.1.0
Restart DSH Web, then click the 🎤 button beside the composer to start continuous recognition. Click it again to stop and insert the transcript into the draft.
zh-CN; accuracy for Chinese, English, and mixed-language speech depends on the browser's recognition service.0.1.0-rc.5 source release.pnpm install
pnpm build
MIT License
CLASSIFICATION EVIDENCE
系统优先读取 GitHub Topics,再与站内分类词典和词根规则比对。