Press your shortcut
Right where you’re writing.
A personal voice layer
Your voice.
Your choice.
Speak naturally. Write anywhere.
Your favorite speech models, one shortcut away.
macOSWindows 11Linux
From a thought to a sentence.Illustration
Wherever you write
EmailMessagesDocumentsCode
One shortcut. That’s it.
Hold. Speak. Done.
Speak your mind
Your words. Your natural pace.
Let go. Keep going.
Text lands when it’s ready.
Fn on Mac. Ctrl+Win on Windows. Your chosen key combination on Linux.
No one-size-fits-all AI
Your favorite model.
At your fingertips.
Cloud
OpenAI
GPT, live or after recording.
LiveAfter recording
Cloud
ElevenLabs
Scribe, with your vocabulary.
LiveAfter recording
On device
Local Whisper
Downloaded once. Runs on your Mac.
OfflinemacOS
Also Groq, OpenRouter, Mistral & compatible endpoints.
A little polish.
Only if you want it.
Use your transcript directly, or add an LLM.
Voice
Speech
Optional LLM
Your text
What’s the difference between live and after recording?
Live models process audio as you speak. Others receive the complete recording after you stop. Both paste the final text when ready, never partial results.
Which models can I use?
OpenAI: gpt-transcribe, gpt-live-transcribe, gpt-4o-mini-transcribe, and whisper-1. ElevenLabs: Scribe v2 and Scribe v2 Realtime. Local Whisper runs offline on macOS after download. Other speech presets include Groq, OpenRouter, and Mistral, with custom compatible endpoints supported. Optional formatting supports OpenAI, Groq, OpenRouter, Mistral, Gemini, Ollama, LM Studio, or a downloaded model on Mac.
Do I need a language model as well?
No. Leave the LLM off for transcription without AI rewriting. Dictionary replacements and basic styles still work. Enable a language model for cleanup or custom writing instructions.
How does the dictionary work?
Names and specialist terms guide recognition; OpenAI Live uses keyword hints. Hints don’t guarantee spelling. Your replacement rules run locally, even with the LLM off. ElevenLabs charges extra for dictionary hints.
Where do my recordings and API keys go?
History stays on your device. Keys are securely stored and remembered per provider. Cloud transcription sends audio to your speech provider; cloud formatting sends text to your LLM provider. For offline dictation on Mac, download local models and keep formatting local or off.
What does it cost?
There’s no app subscription. Bring your own API keys and pay your cloud providers for usage. Downloaded local models have no per-request fees on your Mac.
How do I get started?
Download the latest release for macOS (Apple Silicon), Windows 11 (x64), or Linux (x64, X11 sessions), choose a model, and set your shortcut. The setup guide covers building from source.
How can I contribute?
Report bugs and share ideas in GitHub issues. For code, docs, or translations, follow the setup guide and open a focused pull request with relevant checks.