The best and most complete voice to text for Mac.
50ms latency, 96.5% accuracy, 30+ languages, smart formatting, and custom vocabulary.
100% private: no data leaves your system.
Features
Choose between native zero-config Apple Mode for offline speech recognition, or Advanced Mode to download and cherry-pick local Whisper, Qwen, and Gemma models.
Built-in offline LLM formatting automatically handles punctuation, capitalization, and cleans up filler words — all 100% locally on your Mac.
Slack, VS Code, Cursor, Notion, Gmail, Chrome, Xcode — works in every text field, zero configuration.
Add up to 100 custom keyterms. Pre-built packs for Coding, Medical, Legal, Finance.
Under 50MB RAM. Built with Tauri, not Electron. 10x lighter than every alternative.
No audio uploaded. No transcripts uploaded. No telemetry. Login and subscription data are the only account records.
Modes
Default out-of-the-box mode using system-native, 100% on-device speech-to-text and a built-in offline LLM for cleaning up and formatting transcripts. Zero setup, zero data upload.
Unlock configurable workflows to cherry-pick state-of-the-art on-device models. Choose Speech-to-Text Only (local speech-to-text models), Speech-to-Text + LLM Formatting (local speech model + local formatting LLM), or fully integrated End-to-End local models for your ideal offline pipeline.
How it works
The Mac app is being prepared for launch. Private on-device dictation is the first release.
Coming soonHold one key to talk, release to stop. Or tap to toggle. No training, no calibration.
Text is injected wherever your cursor is. Any app, any field, any language.
Pricing
Coming soon
tapWhisper for Mac is coming soon. Private, on-device voice-to-text is being prepared for launch.
macOS 13+ · Apple Silicon & Intel · Under 20MB app target
Privacy
No audio uploaded. No transcripts uploaded. Models stay local. When accounts launch, the only stored records are login information and subscription data. We do not store audio, transcripts, vocabulary, settings, or model data.
FAQ
tapWhisper offers significantly faster performance compared to slower dictation workflows. Unlike tools that treat offline mode as an afterthought, tapWhisper is built around private on-device transcription with Apple Mode by default and Advanced Mode for custom on-device pipelines.
tapWhisper uses 10x less RAM than Electron-based alternatives thanks to Tauri architecture instead of a bundled browser runtime. You also get custom vocabulary with up to 100 keyterms, and seamless integration with every Mac application without configuration.
Plus, tapWhisper is more affordable at $9.99/month $5.99/month or $99.99/year $49.99/year. Better performance, stronger privacy, lower price.
Yes. tapWhisper uses on-device transcription by default, so it works when you're offline — on airplanes, in areas with poor connectivity, or whenever you want complete privacy. Apple Mode is the default engine, and Advanced Mode is available for custom speech and formatting pipelines.
The transition is designed to feel seamless: you speak, tapWhisper listens, and the text appears in the active app without sending your audio somewhere else first.
Most competitors still assume constant internet access. tapWhisper's local-first architecture means you keep working regardless of connectivity, while your audio and transcripts stay on your Mac.
Absolutely. tapWhisper works system-wide in every text field on macOS — Slack, VS Code, Cursor, Notion, Gmail, Chrome, Xcode, Messages, Notes, and thousands more. Zero configuration required.
Unlike browser-based solutions or app-specific dictation tools, tapWhisper integrates at the system level using macOS accessibility APIs. Press your hotkey anywhere and start speaking — text appears instantly in whatever application you're using.
No app-by-app setup, no browser extension, and no copy-paste round trip. It is designed to disappear into the way you already work.
tapWhisper targets under 50MB of idle RAM — dramatically lighter than Electron-based alternatives. Built with Tauri, tapWhisper uses the native system WebView instead of shipping an entire Chromium runtime with the app.
This means tapWhisper is designed to stay open all day in the menu bar without slowing down your Mac, draining your battery, or interfering with other applications.
Local models are downloaded only when you choose them, and model files stay in the app's local model folder.
tapWhisper supports multilingual speech including English, Spanish, French, German, Italian, Portuguese, Dutch, Russian, Japanese, Korean, Chinese (Mandarin), Arabic, Hindi, and many more. Language detection is automatic where supported, or you can manually select your preferred language.
Switch between languages seamlessly without restarting the app. Apple Mode provides the fast native path, while Advanced Mode allows customizing speech and formatting options.
It is built for people who move between English, native-language notes, names, and specialized vocabulary during the same day.
Yes. tapWhisper is privacy-first by design. Your audio and transcripts stay on your Mac, with no transcript upload, no telemetry SDKs, and no background sync of your dictated text.
Everything stays on your device — no data leaves your Mac. Unlike server-based dictation tools, tapWhisper keeps transcription local and keeps your words under your control.
Your transcripts, custom vocabulary, models, and settings are stored locally. When accounts launch, only login information and subscription data are stored for billing. Nothing else.