TalkType

Stop typing prompts. Just speak.
Your phone's voice input → instant text at your computer's cursor → in any app.
Global voice input for vibe coding.

Quick Start • Features • Comparison • Download

English | 中文

---

⚡ 3x Faster Than Typing

Why type when you can talk?

	Typing	Speaking with TalkType
Speed	~40 WPM	~150 WPM
200-word prompt	~5 minutes	~1.3 minutes
Hands	Both occupied	Free to browse code
Fatigue	RSI risk	Relaxed

Speaking is 3.75x faster than typing. A complex 200-word prompt that takes 5 minutes to type? Speak it in under 80 seconds.

Demo

📹 GIF coming soon — showing: open app → scan QR → speak on phone → text appears in Cursor

Help wanted: If you use TalkType, record a short demo and submit a PR!

🚀 Quick Start

Get started in 30 seconds:

1. Download → Latest Release (macOS / Windows / Linux)
2. Open TalkType on your computer → QR code appears
3. Scan the QR code with your phone → web page opens (no app install)
4. Speak on your phone → tap Send → text appears at your cursor ✨

⚠️ Phone and computer must be on the same WiFi.

Pro Tips

• 📌 Add to Home Screen — on your phone's browser, "Add to Home Screen" for instant access (no rescan needed next time)
• 🌍 Bilingual UI — switch between English and Chinese in the app settings

✨ Features

• 🎙️ Phone-native voice input — uses your phone's built-in STT, no model download needed
• ⌨️ Global text injection — text appears at your cursor in ANY app
• 📱 Zero install on phone — phone client is a web page, no app needed
• 🔒 100% local network — all data stays on your WiFi, zero cloud dependency
• 📷 QR code pairing — scan and connect in seconds
• ↵ Auto-Enter — optionally press Enter after sending, perfect for chat apps
• 🔁 Auto reconnect — remembers connections, auto-reconnects on same WiFi
• 🌍 i18n — English and Chinese UI, auto-detects your language
• 🌍 Cross-platform — macOS, Windows, Linux
• 📲 PWA support — add to phone home screen for app-like experience

🎯 Why TalkType?

The Unfair Advantage

Other voice input tools run Whisper locally (1–3 GB model download, CPU/GPU intensive) or use cloud APIs (latency, cost, privacy risk).

TalkType takes a fundamentally different approach: your phone IS the voice engine. Apple's, Google's, and Samsung's STT are trained on billions of data points, support 100+ languages, and are already on your device. TalkType simply bridges that to your computer over local WiFi.

Zero config. Zero cost. Zero compromise on privacy.

vs. Other Voice Input Tools

	TalkType	SuperWhisper	Wispr Flow	VoiceTypr	MacWhisper
Price	Free	$9/mo	$15/mo	Free	Freemium
Open Source	✅	❌	❌	✅	❌
Voice Engine	Phone native	Local Whisper	Cloud AI	Local Whisper	Local Whisper
Model Download	None	1–3 GB	N/A	1–3 GB	1–3 GB
Setup Time	30 sec	5–10 min	5 min	5–10 min	5 min
Languages	100+ native	~50	~30	~50	~50
macOS	✅	✅	✅	✅	✅
Windows	✅	❌	❌	✅	❌
Linux	✅	❌	❌	❌	❌
Privacy	LAN only	Local	☁️ Cloud	Local	Local

💡 Use Cases

Scenario	Description
🧑‍💻 Vibe Coding	Speak prompts to Cursor, Copilot, Windsurf, or any AI coding tool
🤖 AI Chat	Dictate to ChatGPT, Claude, Gemini — 3x faster than typing
📝 Writing	Draft emails, docs, and notes hands-free
💬 Messaging	Reply in Slack, Discord, WeChat, Teams by voice
🔍 Search	Voice search in any app or browser

📦 Download

Go to Releases and download for your platform:

Platform	File	Notes
🍎 macOS (Apple Silicon)	TalkType_x.x.x_aarch64.dmg	M1 / M2 / M3 / M4
🍎 macOS (Intel)	TalkType_x.x.x_x64.dmg	Older Intel Macs
🪟 Windows	TalkType_x.x.x_x64-setup.exe	Windows 10 / 11
🐧 Linux	.deb / .AppImage	Ubuntu, Debian, etc.

💡 Which Mac? Click → About This Mac → check "Chip". Apple M1/M2/… = aarch64, Intel = x64.

⚙️ Platform Notes

🍎 macOS

• Grant Accessibility permission on first launch: System Settings → Privacy & Security → Accessibility → Enable TalkType ✅
• The app guides you through this on first run
• Closing the window minimizes to menu bar tray (right-click → Quit)

🪟 Windows

• Allow firewall access when prompted (required for phone to connect)
• Closing the window minimizes to system tray (right-click → Quit)

🐧 Linux

• .deb package handles dependencies automatically
• For .AppImage: run chmod +x TalkType_*.AppImage first

❓ FAQ

Why use phone voice input instead of local Whisper?

Phone STT engines (Apple, Google, Samsung) are trained on billions of data points, support 100+ languages natively, and are heavily optimized for mobile hardware. No model download (1–3 GB saved), no GPU required, no configuration — it just works, instantly.

Is my data safe?

Yes. All communication happens over your local WiFi via WebSocket. No data ever leaves your network. No cloud, no accounts, no telemetry.

Phone can't open the page after scanning?

• Ensure phone and computer are on the same WiFi
• Check your computer's firewall settings
• Try disabling VPN or proxy

Connected but no text appearing?

• macOS: Check Accessibility permission (System Settings → Privacy & Security → Accessibility)
• Make sure an input field is focused on your computer
• Text goes to wherever your cursor currently is

🔧 How It Works

┌─────────────┐       WiFi (LAN)        ┌──────────────┐
│  📱 Phone    │  ◄── WebSocket ──►      │  💻 Computer  │
│  (Browser)   │      tap to send        │  (Desktop App)│
│              │                         │       ↓       │
│  🎤 Voice    │                         │  ⌨️ Keyboard  │
│  → Text      │                         │   Injection   │
└─────────────┘                          └───────┬───────┘
                                                 ↓
                                          Any Active App
                                        (Cursor / Claude /
                                         ChatGPT / Docs)

Tech Stack:

• Desktop: Tauri 2 + Rust (Axum HTTP/WS server + keyboard simulation)
• Mobile: Pure HTML/CSS/JS PWA (served by desktop app, zero install)

🛠️ Development

Build from source

Prerequisites

• Rust 1.70+ (latest stable recommended)
• Node.js 18+
• Tauri CLI: cargo install tauri-cli --version "^2"

Dev Mode

git clone https://github.com/puke3615/TalkType.git
cd TalkType
npm install
cargo tauri dev

Build

cargo tauri build

Output: src-tauri/target/release/bundle/

Project Structure

TalkType/
├── src/index.html              # Desktop UI (i18n, QR code)
├── src-tauri/
│   ├── src/
│   │   ├── lib.rs              # App entry & Tauri commands
│   │   ├── server.rs           # HTTP + WebSocket server (Axum)
│   │   ├── mobile.html         # Mobile PWA page (i18n, served at /)
│   │   ├── keyboard.rs         # Keyboard simulation (per-platform)
│   │   ├── network.rs          # LAN IP discovery
│   │   └── qrcode_gen.rs       # QR code generation
│   └── tauri.conf.json
└── .github/workflows/          # CI/CD (macOS + Windows + Linux)

📄 License

MIT — free to use, modify, and distribute.

---

If TalkType saves you time, give it a ⭐!
_{Built with ❤️ using Tauri 2 + Rust}