SayType

English · 中文说明 ↓

A voice input method built with Tauri that lets you dictate text in any application using a global hotkey and AI transcription.

Features

• Hold-to-Record Hotkey: Hold down Ctrl+Shift to start recording, release to stop and transcribe
• Real-time Audio Visualization: Waveform animation while recording
• AI Transcription: Cloud (Groq / OpenAI Whisper API) or a fully local engine (Qwen3-ASR, ~1 GB one-time download, no API key, audio never leaves your machine)
• Auto-typing: Inserts transcribed text into the active application (macOS; Windows/Linux experimental, untested)
• Background Operation: Runs silently in the system tray
• Automatic Updates: New releases download in the background; restart from the tray when you're ready
• Customizable Settings: Configure API key/provider, microphone, and language

Requirements

• Node.js 16 or higher
• A Rust toolchain (rustup) — required to build the Tauri app
• A Groq or OpenAI API key — only for the cloud providers; the local engine needs no key (recommended on Apple Silicon)
• Microphone access permission (and Accessibility permission on macOS)

Download

Grab the latest installer from the Releases page:

• macOS — the .dmg
• Windows — the _x64-setup.exe (or .msi)
• Linux — the .AppImage (portable), .deb (Debian/Ubuntu), or .rpm (Fedora/RHEL)

Once installed, SayType keeps itself up to date. The other files on the Releases page (*.sig, latest.json, *.app.tar.gz) are used by the built-in auto-updater — you don't need to download them.

Installation / Development

# Install JS tooling (only @tauri-apps/cli)
npm install

# Run in development mode
npm run dev        # = tauri dev

# Build for production
npm run build      # current host target
npm run build:mac  # macOS (aarch64)
npm run build:win  # Windows
npm run build:linux

Then open Settings to configure your API key.

To publish a signed macOS .dmg to GitHub Releases (built automatically on a v* tag), see RELEASING.md.

Usage

1. Launch SayType
2. Pick a transcription engine in Settings: download the local model, or enter a Groq/OpenAI API key
3. Hold down Ctrl+Shift to start recording
4. Speak while holding the keys
5. Release to stop recording and transcribe
6. Text is inserted into the active application (macOS; Windows/Linux experimental, untested)
7. Press Escape to cancel recording or an in-progress transcription

Configuration

Access settings through the tray menu or main window to configure:

• Transcription engine: local model (no key) or cloud provider (Groq / OpenAI) with API key
• Default microphone
• Transcription language and custom dictionary (cloud providers only; the local engine auto-detects)

Reset macOS permissions for repeated testing

tccutil reset Accessibility com.tao.saytype
tccutil reset Microphone com.tao.saytype

License

PolyForm Noncommercial 1.0.0 https://polyformproject.org/licenses/noncommercial/1.0.0/

---

SayType(中文)

↑ English

一款基于 Tauri 构建的语音输入法,让你通过全局快捷键和 AI 转写,在任意应用中用口述输入文字。

功能特性

• 按住录音快捷键:按住 Ctrl+Shift 开始录音,松开即停止并转写
• 实时音频可视化:录音时显示波形动画
• AI 转写:云端(Groq / OpenAI Whisper API)或完全本地引擎(Qwen3-ASR,一次性下载约 1 GB,无需 API key,音频不出本机)
• 自动输入:将转写结果插入到当前活动应用(macOS;Windows/Linux 为实验性支持,未经真机验证)
• 后台运行:静默驻留在系统托盘
• 自动更新:新版本后台自动下载,你随时从托盘重启完成升级
• 可自定义设置:配置 API key / 服务商、麦克风和语言

环境要求

• Node.js 16 或更高版本
• Rust 工具链(rustup)—— 构建 Tauri 应用所需
• Groq 或 OpenAI API key——仅云端服务商需要;本地引擎无需 key(Apple Silicon 上推荐)
• 麦克风访问权限(macOS 上还需辅助功能权限)

下载

从 Releases 页面 下载对应平台的安装包:

• macOS — .dmg
• Windows — _x64-setup.exe(或 .msi)
• Linux — .AppImage(免安装)、.deb(Debian/Ubuntu)或 .rpm(Fedora/RHEL)

安装后 SayType 会自动保持更新。Releases 页面上其余文件(*.sig、latest.json、*.app.tar.gz)是自动更新用的,无需下载。

安装 / 开发

# 安装 JS 工具链(仅 @tauri-apps/cli)
npm install

# 以开发模式运行
npm run dev        # = tauri dev

# 构建生产版本
npm run build      # 当前主机架构
npm run build:mac  # macOS(aarch64)
npm run build:win  # Windows
npm run build:linux

然后打开「设置」配置你的 API key。

要把签名后的 macOS .dmg 发布到 GitHub Releases(打 v* tag 时自动构建),见 RELEASING.md。

使用方法

1. 启动 SayType
2. 在「设置」中选择转写引擎:下载本地模型,或填入 Groq/OpenAI API key
3. 按住 Ctrl+Shift 开始录音
4. 按住按键的同时说话
5. 松开按键停止录音并转写
6. 文字会插入到当前活动应用(macOS;Windows/Linux 为实验性支持,未经真机验证)
7. 按 Escape 取消录音或正在进行的转写

配置

通过托盘菜单或主窗口进入「设置」,可配置:

• 转写引擎:本地模型(无需 key)或云端服务商(Groq / OpenAI)+ API key
• 默认麦克风
• 转写语言和自定义词典(仅云端服务商;本地引擎为自动检测)

重复测试时重置 macOS 权限

tccutil reset Accessibility com.tao.saytype
tccutil reset Microphone com.tao.saytype

许可证

PolyForm Noncommercial 1.0.0 https://polyformproject.org/licenses/noncommercial/1.0.0/