Evidence snapshot reviewed Sep 10, 2026GitHub checked Aug 21, 2026
DSHub.
Plugin RSS
Evidence-verifiedPlugin BundleInterface & ExperienceWeb Profile

DSH Voice Scribe

Dictate into the DSH web composer with Alt-key or microphone controls.

Install this Plugin ↓View on GitHub ↗More Plugin Bundles

At a glance

What it does

Dictate into the DSH web composer with Alt-key or microphone controls.

Use cases
Interface & ExperienceMedia CaptureUIIntegrations
Works with
WebMicrophoneOptional Cloud AsrOptional Llm Provider
Compatibility

Web Profile
>=0.1.0-rc.6; declared pre-release ranges in package manifest

Trust & status

Evidence-verified
Checked Sep 9, 2026, 2:22 PM UTC

Code-evidenced contributions

What it adds to DSH

Web UIVoice input for the DSH composer

Adds microphone controls and Alt-key dictation that appends transcribed speech to the composer draft.

Mechanism evidence ↗

Before you choose it

DSH Voice Scribe is a DSH web plugin for adding speech-to-text to the composer. Tap the microphone or use Alt to record, then append the transcription to the existing draft. Its automatic mode prefers local SenseVoice recognition and falls back to browser Web Speech; optional OpenAI-compatible cloud ASR and DSH-model polishing are available.

Best for

DSH web users who want hands-free drafting, especially those who prefer a local transcription option before using browser or cloud speech services.

Common tasks

  • Dictate a prompt or chat draft without overwriting text already in the composer.
  • Use local offline SenseVoice transcription with automatic language detection.
  • Correct recurring names or project terms with the optional hot-word replacement file.
  • Optionally polish a transcription through an already configured DSH model.

Permissions and data

Voice recording is required for transcription. Handling depends on the engine selected.

Permissions
  • Microphone access is needed to record speech.
  • Optional cloud ASR requires configuring an API key.
Data handling
  • The README states that local SenseVoice audio stays on the device.
  • Browser Web Speech is handled by browser speech services.
  • Cloud ASR can send audio to the configured external endpoint.
  • Cloud ASR keys are stated to be stored server-side.
  • Custom polishing prompts are stated to be stored server-side.
  • Optional hot-word rules are read from `$DSH_HOME/voice/hot.txt`.
External services
  • Browser Web Speech services may be used when the local engine is unavailable.
  • Optional OpenAI-compatible cloud ASR endpoints may be configured.
  • Optional polishing reuses a configured DSH model.
Credentials
  • No API key is described for local or browser recognition.
  • An API key is required only when configuring cloud ASR.

Limitations

  • Browser Web Speech depends on external speech services and may be unavailable in some networks or browser versions.
  • With `@` reference chips in a draft, inserting a transcription can expand chips into plain text.
  • The local model download is approximately 230 MB on first use.
  • The listed recording limits are about four minutes for local recognition and ten minutes for cloud recognition.

What DSHub checked

  • Pinned Git source and DSH bundle structure were verified.
  • The package declares Node.js 18+ and a DSH LLM peer-dependency range.
  • The README documents DSH web installation, voice controls, recognition options, and stated compatibility.

What DSHub did not check

  • Installation and microphone behavior were not executed during this curation.
  • Actual recognition accuracy, model download behavior, network availability, and cloud-ASR failover were not independently tested.
  • Published npm package contents were not audited.

Pinned install

Install DSH Voice Scribe

This plugin bundle does not have a DSH Plugin install action. Use its source documentation for the delivery method.

Visit the source project ↗

Maintainer source

Project README

View at commit 7a6172a ↗
Maintainer-authored contentCaptured from README.md on Sep 9, 2026. The text and repository-relative media are fixed to commit 7a6172aedd1f with content hash 6e6a4f5c4914; provider-hosted badges may update independently. README commands are upstream documentation; the DSHub copy action above is the verified, version-pinned install.

dsh-voice-scribe

MIT license release CI npm version npm downloads dsh.so risk dsh.so install Listed in awesome-dsh-plugin

DSH 专属语音输入插件：点按或按住 Alt 说话、松开/再点按转文字，结果追加到输入框草稿末尾（不覆盖已输入内容）。 Voice input for DeepSeek Harness: tap or hold Alt to talk, get text in the composer.

⚠️ 非官方插件，与 DeepSeek / 深度求索公司无关联。使用前请阅读 SECURITY.md。

安装 Install

dsh plugin --profile web add dsh-voice-scribe   # 重启 dsh web 后生效

使用 Usage

  • 麦克风按钮：输入框右侧 🎤 图标，点击开始说话、再点停止并转写（按钮录音中变红）
  • 热键：点输入框 → 按 Alt 开始说话 → 再按 Alt 结束并转写（备选 Alt+空格，设置可切换）
  • 按住说话：设置 → 语音输入 → 触发方式 可选「按住说话」——按住热键录音、松开自动转写（麦克风按钮同样支持）
  • 实时中间结果：说话时识别文本实时出现在草稿里（浏览器引擎逐字、本地引擎每 3 秒刷新），停止后替换为最终结果
  • 录音电平指示：录音中状态条下方显示实时电平条
  • 最长录音时长：本地引擎约 4 分钟、云端 10 分钟，到时长自动停止并转写
  • 切窗取消：录音中切到其他窗口自动取消本次录音（Alt+Tab 误触不会留下录音）

识别引擎 Engine（默认「自动」，零配置）

引擎 说明
自动（默认） 本地离线识别优先；不可用时自动回退浏览器识别
本地离线识别 SenseVoice，零配置零 key、音频不出本机；首次使用自动下载模型（约 230MB，国内镜像，只需一次）
浏览器 Web Speech 零配置；依赖 Google/Microsoft 服务（国内 / Edge Stable 可能不可用）
云端 ASR（可选） 服务链：可配置多个 OpenAI 兼容端点按序尝试、失败自动切换；需在设置中配置 API key

浏览器识别依赖外部语音服务（Chrome 在大陆被墙、Edge Stable 有已知回归），故默认以本地识别为主。

云端 ASR 服务链示例：Groq（免费层）→ 硅基流动 SenseVoice → 阿里云百炼，任一失败自动尝试下一个（设置 → 语音输入 → 云端 ASR）。

识别语言 Languages

支持 中文 / English / 粤语 / 日本語 / 한국어（设置 → 语音输入 可选）。本地离线识别自动检测语言；所选语言作用于浏览器与云端识别。

热词替换表 Hot Words（可选）

把识别错的人名、术语、项目名替换回来：编辑 $DSH_HOME/voice/hot.txt（每行一条，修改后下次转写生效）：

# 字面替换（不区分大小写）：正确词=错误词1|错误词2
DeepSeek=deep seek|迪普西克
王小明=王小铭

# 正则替换（标准 $1 语义）
/老\s*师/老师/
/\{([^}]+)\}/【$1】/

设置 → 语音输入 页面会显示热词表状态（规则条数 / 文件路径 / 解析错误）。云端与本地离线引擎的转写结果统一应用。

自定义润色提示词（可选）

设置 → 语音输入 → 开启润色后：

  • 润色模型：下拉选择复用的 DSH 模型（选项来自 DSH 已配置的 provider，首次开启自动选中第一个）
  • 润色提示词：可自定义（多行，保存在服务端）；留空或「恢复默认」使用内置的最小必要修正提示词

润色时会先做一步本地规则预润色（去「嗯/呃」等口头禅、折叠多余空格），再把更短更干净的文本交给 LLM，省 token；LLM 失败时仍保留原始转写。

隐私 Privacy

本地引擎音频不出本机；Web Speech 由浏览器语音服务处理；云端 ASR 的 key 只存服务端。

与同类插件对比 Compare

同为 DSH 的语音 / 输入增强插件，主要差异（截至 2026-09）：

dsh-voice-scribe（本插件） dsh-better-input
定位 专注语音输入 输入增强套件（语音 + 提示词优化 + 文件转 Markdown 等）
本地离线识别 ✅ SenseVoice，零 key，音频不出本机 ❌ 仅浏览器原生识别
浏览器 Web Speech ✅ 回退 ✅
云端 ASR 服务链 ✅ 多 provider 故障切换 ❌
热词替换表 hot.txt ✅ ❌
本地规则预润色（省 token） ✅ 0.4.2 起 ❌
AI 润色（复用 DSH 模型） ✅ ✅
按住说话 / 录音电平 ✅ 录音自动停止（无电平）
提示词优化 / 文件转 Markdown ❌ ✅

只想要更省心、更私密的语音输入 → dsh-voice-scribe；需要一整套输入增强（提示词优化、文件转 Markdown） → dsh-better-input。两者可并存。

已知限制 Known limitations

  • 输入框里含 @ 引用芯片（如 @文件）时，DSH 只提供「整段替换草稿」的接口，转写结果插入会把芯片展开成纯文本；先发送或清空草稿再听写可避免。
  • 浏览器 Web Speech 依赖外部语音服务，国内网络下通常需要改用本地离线或云端引擎。

兼容性 Compatibility

  • 需要 DSH 0.1.0-rc.6 及以上（含 0.1.1-rc / 0.1.2-alpha / 0.1.3-alpha / 0.1.5-alpha 各预发布线）。
  • 输入框插槽 conversation.input.right 在 DSH 0.1.2 起由 <textarea> 改为 Lexical contenteditable：0.4.8 起两种形态都支持（读取实时草稿走 useInput，写入走 inputActions.setDraft）。
  • 界面没有麦克风按钮（旧壳子没有该插槽）时，Alt 热键仍然可用。

开发 Dev

npm test          # 测试
npm run lint      # 语法检查
npm run security  # 密钥/路径泄露扫描

License

MIT

Operate deliberately

Install and manage

Prerequisites and target Profile

Target： Web Profile

Delivery： Dsh Bundle Git — PensiveFei/dsh-voice-scribe#7a6172aedd1fe7ecfdd4217438b9a3f0e394287b。

Verify, update, and remove

Show lifecycle commands
Verify
dsh plugin --profile web list

Compatibility and access

DSH web bundle; Node.js 18+ declared： >=0.1.0-rc.6; declared pre-release ranges in package manifest。

Review compatibility evidence ↗

Risk facts

Microphone

Uses microphone recording for voice transcription.

Evidence ↗
External Services

Browser speech recognition and optional cloud ASR may use external speech services; optional cloud ASR requires an API key.

Evidence ↗
Local Model Download

The local recognition engine may download a roughly 230 MB model on first use.

Evidence ↗
Evidence and editorial reviewManifest, Bundle patch, distribution and freshness

Immutable evidence

Review status and source activity

AI reviewed

This is an unofficial plugin and is not affiliated with DeepSeek, as stated by its README.

AI reviewed Sep 10, 2026, 11:36 AM UTC。GitHub facts last checked Sep 10, 2026, 11:36 AM UTC。

No material source change has been recorded since this evidence baseline.

Next step

Follow the Plugin installation workflow →

Subscribe to material changes for DSH Voice Scribe →