tover0314-w/opentypeless

English | 中文 | 日本語 | 한국어 | Español | Français | Deutsch | Português | Русский | العربية | हिन्दी | Italiano | Türkçe | Tiếng Việt | ภาษาไทย | Bahasa Indonesia | Polski | Nederlands

Open-source alternative to Wispr Flow and Superwhisper for AI voice input, rewriting, and voice Q&A on macOS, Windows, and Linux.

Press a hotkey, speak naturally, and get clean text in the app you are already using.
Or ask a one-shot voice question and get a concise AI answer without opening a chat app.

Website · Download · Run locally · Discussions

If OpenTypeless helps your workflow, a GitHub star helps more people discover the project.

Dictate anywhere · Rewrite selected text · Ask one-shot voice questions · Bring your own keys or use managed cloud words

• Stripe checkout for new purchases with existing Creem subscriptions kept valid.
• More reliable cloud sign-in and billing with secure session storage, clearer service-status messages, and stronger account-state recovery.
• Faster and more reliable providers through connection reuse and warm-up, complete Deepgram and AssemblyAI final transcripts, and preserved Custom Whisper URL queries.
• Modern model support for GPT-5/o-series requests and the native Anthropic Claude Messages API, while Claude through OpenRouter remains supported.
• More STT choices and safer recording limits including regional Qwen3 ASR Realtime, provider-aware duration controls, and preserved audio while providers connect.
• Better platform support including Linux ARM64 packages, Wayland direct typing with compositor-specific helpers, and safer Windows modifier-key output.

See the complete notes and verified installers on the latest release page.

OpenTypeless understands more of the workflow around your voice, not only the words you said:

• App-aware writing detects the active application locally and adapts structure and tone for email, chat, documents, issue trackers, coding tools, and more.
• Voice intent routing distinguishes dictation, selected-text editing, translation, Ask Anything, and supported voice actions in English, Simplified Chinese, and Traditional Chinese.
• Multiple shortcuts per workflow let you assign and reorder more than one binding for Dictation, Ask Anything, and Translation.
• Switchable translation targets make it faster to move between the languages you use instead of keeping one fixed output language.
• A stronger local dictionary adds correction rules plus dictionary import/export, so recurring names, terminology, and transcription mistakes stay under your control.
• Per-app style mappings let you override the built-in app category when a detected app needs a different writing style.

All application detection, mappings, dictionary entries, and correction rules are stored locally. App-aware polish sends only the internal app category and approved style metadata to the configured LLM path; raw window titles and document contents are not sent to the LLM or stored in history.

Ask Anything is a shortcut-first voice Q&A flow, not a chat tab. Press the Ask Anything hotkey, speak a question, stop recording, and OpenTypeless transcribes it, sends a one-shot request to the LLM, then shows only the final answer in a small floating note.

It is designed for quick answers with no chat history, no input box, and no extra send step. If selected-text context is enabled, Ask can also answer questions about the text currently selected in another app.

Default shortcuts now follow the Typeless-style flow:

Linux keeps conservative Ctrl-based defaults because global Right Alt handling is less reliable across desktop environments, especially on Wayland.

App-aware AI polish	Local dictionary and corrections

Onboarding and voice workspace

Onboarding	Voice workspace

OpenTypeless gives you four voice-first desktop workflows:

Use it for emails, chat replies, meeting notes, issue comments, prompts, documentation drafts, quick answers, multilingual translation, and any workflow where speaking is faster than typing.

flowchart LR
  Hotkey["Global hotkey"] --> Capture["Audio capture"]
  Capture --> Mode{"Voice mode"}
  Mode -->|"Dictation"| STT["Speech-to-text"]
  STT --> Polish{"AI polish?"}
  Polish -->|"Yes"| Context["Local app category + style mapping"]
  Context --> LLM["Context-aware LLM rewrite"]
  Polish -->|"No"| Output["Keyboard / clipboard output"]
  LLM --> Output
  Mode -->|"Ask Anything"| AskSTT["Transcribe question"]
  AskSTT --> AskLLM["One-shot LLM answer"]
  AskLLM --> Panel["Answer-only panel"]
  Mode -->|"Selected text"| Selection["Capture selected text"]
  Selection --> EditLLM["Apply spoken instruction"]
  EditLLM --> Output

Most desktop dictation tools stop at transcription. OpenTypeless adds the AI rewrite layer, provider choice, and open-source control that power users need.

UI localization currently ships with complete English and Chinese copy, plus additional locale files that may still fall back to English for newer advanced features.

sequenceDiagram
  participant User
  participant Desktop as OpenTypeless Desktop
  participant STT as STT Provider
  participant LLM as LLM Provider
  participant App as Active App

  User->>Desktop: Hold dictation hotkey and speak
  Desktop->>Desktop: Detect local app category and style
  Desktop->>STT: Send audio via BYOK or Cloud
  STT-->>Desktop: Raw transcript
  Desktop->>LLM: Optional context-aware polish / translate / rewrite
  LLM-->>Desktop: Final text
  Desktop->>App: Type or copy result

1. Download the latest build for your platform from Releases.
2. Choose BYOK for full provider control or Cloud if you want managed quota without API keys.
3. Pick speech recognition and AI polish providers in Settings.
4. Set one or more shortcuts for Dictation, Ask Anything, and Translation.
5. Open any desktop app, press the hotkey, speak, and let OpenTypeless type the polished result.

Download the latest version for your platform:

Download the latest release

Release signing differs by platform while distribution is being improved. Always download from the official GitHub Releases page.

Windows SmartScreen may show "Windows protected your PC":

1. Click More info
2. Click Run anyway

If the installer shows a publisher validation warning:

1. Right-click the .msi file → Properties
2. Check Unblock at the bottom → Apply
3. Run the installer again

macOS builds are Developer ID signed. If Gatekeeper still blocks first launch while notarization/stapling catches up, remove the quarantine attribute:

xattr -cr /Applications/OpenTypeless.app

Then open the app normally.

Ubuntu/Debian — install the .deb package:

sudo apt install ./OpenTypeless_x.x.x_amd64.deb

AppImage — make it executable and run:

chmod +x OpenTypeless_x.x.x_amd64.AppImage
./OpenTypeless_x.x.x_amd64.AppImage

NVIDIA + Wayland users: The app auto-detects this configuration and applies a workaround. If it still crashes on startup, run:

WEBKIT_DISABLE_DMABUF_RENDERER=1 ./OpenTypeless

Wayland users: the required helper depends on the compositor:

• KDE Plasma/KWin Wayland uses kwtype, which must currently be built and installed using the upstream instructions.
• Other Wayland compositors use wtype. On Ubuntu/Debian run sudo apt install wtype; on Arch/Manjaro run sudo pacman -S wtype.

Global hotkeys can still depend on the desktop environment. If the matching helper is unavailable or fails, OpenTypeless keeps the result in the clipboard instead of running an untrusted user script.

• Node.js 20+
• Rust (stable toolchain)
• Platform-specific dependencies for Tauri: see Tauri Prerequisites

# Install dependencies
npm install

# Run in development mode
npm run tauri dev

# Build for production
npm run tauri build

The built application will be in src-tauri/target/release/bundle/.

All settings are accessible from the in-app Settings panel:

• Speech Recognition — choose STT provider and enter your API key
• AI Polish — choose an LLM provider and model, enable app-aware writing, manage per-app style mappings, and configure selected-text context
• General — manage multiple Dictation, Ask Anything, and Translation shortcuts, output mode, auto-start, and idle capsule visibility
• Dictionary — add or import custom terms, export your dictionary, and create local correction rules for recurring transcription mistakes
• Scenes — built-in and local prompt templates with import/export for reusable writing styles
• Account / Upgrade — sign in, check cloud words, manage Pro or Lifetime Starter access

API keys are stored locally in the OS credential vault where available, with a local fallback for unsupported environments. No BYOK keys are sent to OpenTypeless servers — STT/LLM requests go directly to the provider you configure.

OpenTypeless also offers optional managed cloud access so you do not need your own provider keys. Pro and Lifetime Starter plans include shared cloud words for speech recognition and AI rewriting. BYOK remains fully supported.

Learn more about Pro

All core features — recording, transcription, AI polish, keyboard/clipboard output, dictionary, history — work entirely offline from OpenTypeless servers in BYOK mode.

To run OpenTypeless without any cloud dependency:

1. Choose any non-Cloud STT and LLM provider in Settings
2. Enter your own API keys
3. That's it — no account or internet connection to www.opentypeless.com is needed

If you want to point the optional cloud features at your own backend, set these environment variables before building:

# Example: build with a custom backend
VITE_API_BASE_URL=https://my-server.example.com API_BASE_URL=https://my-server.example.com npm run tauri build

Desktop pipeline:

flowchart TB
  UI["React + TypeScript UI"] --> Tauri["Tauri commands"]
  Hotkeys["Global shortcuts"] --> Tauri
  Tauri --> Audio["Rust audio capture"]
  Audio --> Providers["STT providers"]
  Providers --> Pipeline["Pipeline orchestration"]
  Pipeline --> LLMs["LLM providers"]
  Pipeline --> Storage["SQLite history + dictionary"]
  Pipeline --> Output["Keyboard / clipboard output"]
  Account["Account + updater + cloud quota"] --> UI
  Account --> Tauri

src/                  # React frontend (TypeScript)
├── components/       # UI components (Settings, History, Capsule, etc.)
├── hooks/            # React hooks (recording, theme, Tauri events)
├── lib/              # Utilities (API client, router, constants)
└── stores/           # Zustand state management

src-tauri/src/        # Rust backend
├── audio/            # Audio capture via cpal
├── stt/              # STT providers (Deepgram, AssemblyAI, Whisper-compat, Cloud)
├── llm/              # LLM providers (OpenAI-compat, Cloud)
├── output/           # Text output (keyboard simulation, clipboard paste)
├── storage/          # Config (tauri-plugin-store) + history/dictionary (SQLite)
├── app_detector/     # Detect active application for context
├── pipeline.rs       # Recording → STT → LLM → Output orchestration
└── lib.rs            # Tauri app setup, commands, hotkey handling

• Usage summary UI for aggregate audio time and word counts
• More provider-specific setup diagnostics
• Better Linux desktop-environment guidance
• More workflow presets for writing, coding, and support replies
• Plugin-style provider extensions

Is my audio sent to the cloud? In BYOK mode, audio goes directly to your chosen STT provider or local endpoint. Nothing passes through OpenTypeless servers. In Cloud mode, audio is sent to the managed proxy for transcription and quota accounting.

Can I use it offline? With a local Whisper-compatible STT endpoint and a local OpenAI-compatible LLM such as Ollama, the app can run without OpenTypeless cloud services.

Which languages are supported? STT supports 99+ languages depending on the provider. AI polish and translation support 20+ target languages.

Is the app free? Yes. The app is fully functional with your own API keys (BYOK). Cloud plans are optional.

• 💬 Discord — Chat, get help, share feedback
• 🗣️ GitHub Discussions — Feature proposals, Q&A
• 🐛 Issue Tracker — Bug reports and feature requests
• 📖 Contributing Guide — Development setup and guidelines
• 🔒 Security Policy — Report vulnerabilities responsibly
• 🧭 Vision — Project principles and roadmap direction

Contributions are welcome! See CONTRIBUTING.md for development setup and guidelines.

Looking for a place to start? Check out issues labeled good first issue.

• Tauri for the desktop shell
• React and TypeScript for the UI
• Rust for audio capture, providers, hotkeys, output, and local storage
• i18next for multilingual UI

MIT