A thought worth keeping
Transcription history keeps past recordings close. Review the words, replay the audio, or copy a thought into something new.
Your transcription historyThink it. Yap it.
Free, local-first dictation.
Open source. Yours to keep.
Ready when you are400 words. Nearly 6 minutes to type.
3 minutes to say with Yap.
macOS · Windows · Linux
One shortcut.
Back to your flow.
Keep the app you’re in.
Lose the typing detour.
Right ⌥
Your recording key on Mac.
Set the shortcut that suits you.
- Press & speak
Start recording from wherever you’re working. Get the whole thought out.
- Let Yap transcribe
Your chosen Whisper model turns the recording into text.
- Carry on
Auto-paste puts your words into the focused app. No copy-paste detour.
Your voice.
Your choice.
Stays on your machine.
Download a Whisper model once. Then transcribe offline, with your audio processed locally by whisper.cpp.
- Audio processing
- On-device
- After model download
- Works offline
- Transcription fees
- None
Cloud, on your terms.
Connect your own OpenAI API key. Audio is sent to OpenAI for transcription when you choose cloud mode.
- Audio processing
- OpenAI API
- Connection
- Internet required
- Transcription fees
- Your OpenAI usage
Small download.
Or big ambitions.
Pick the model for your machine.
Change it when your needs change.
~75 MBapproximate download
Whisper Tiny
Quick notes and lighter machines. The smallest download and the fastest starting point.
- Speed
- Fastest
- Accuracy
- Basic
~150 MBapproximate download
Whisper Base
A good place to start. Everyday dictation without a large download. Try base.en if you work in English.
- Speed
- Fast
- Accuracy
- Good
~500 MBapproximate download
Whisper Small
A little more room for accuracy. A useful middle ground for everyday, multilingual dictation.
- Speed
- Balanced
- Accuracy
- Better
~1.5 GBapproximate download
Whisper Medium
More accuracy, more computation. A fit when you can trade some speed for a larger model.
- Speed
- Slower
- Accuracy
- High
~3 GBapproximate download
Whisper Large v3
The largest model in the lineup. Choose it when accuracy matters more than download size or processing speed.
- Speed
- Slowest
- Accuracy
- Highest
Relative model trade-offs, not benchmarks. Performance depends on your hardware and language.
Little things.
Less friction.
Out of your way
A compact recording overlay and global hotkeys let you dictate without switching your attention to another window.
Recording with YapOpen all the way down
Free to use. MIT licensed. Built with Go, React, and whisper.cpp. Read the code, build it yourself, or help make it better.
Explore the source on GitHubBefore you Yap.
Is it really free?
Yes. Yap is free and open source, and local transcription has no usage fees. Optional cloud transcription uses your own OpenAI API key and is billed by OpenAI.
Does my voice leave my computer?
Not during local transcription. After downloading a model, you can transcribe offline. If you select OpenAI cloud mode, your audio is sent to OpenAI for transcription.
Which computers can run it?
Yap has macOS, Windows, and Linux downloads. Model performance depends on your hardware. Check the download options and installation guide for your platform.
Can I dictate into my coding tools?
Yap pastes transcribed text into the focused application. You can use it for a prompt, an email, or a note. It supplies the words; your coding tool handles the code.
Got a thought?
Give it a voice.
Download Yap
Free. Open source.
macOS, Windows & Linux.