Whisper Notes now supports Whisper Large-v3 Turbo. Same transcription accuracy, 5× faster. Currently available on Mac with Apple Silicon.
What's Different
1. 5× Faster
Turbo processes audio five times faster than the standard Whisper model. In practice:
- •iPhone 15 Pro: 10 minutes of audio transcribes in ~82 seconds (was 425)
- •MacBook M2: 10 minutes in ~63 seconds (was 316)
- •60% less battery drain
2. Same Accuracy
Turbo doesn't sacrifice quality for speed:
- •Identical accuracy to the standard model
- •Same timestamp precision
- •All 100+ languages supported—automatic language detection works the same
3. Still 100% Offline
Turbo runs on your device just like the standard model:
- •Everything processes locally—your audio never leaves your device
- •Zero server communication
- •Works completely offline
Device Requirements
Currently, Whisper Large-v3 Turbo is available on Mac with Apple Silicon (M1/M2/M3/M4 or newer).
We're closely monitoring future model updates for iPhone and iPad support. As AI models become more efficient, we hope to bring Turbo to mobile devices.
Why the limitation? Whisper Notes is 100% offline—users can't download models after installing the app. This means we ship one model that must balance performance across mainstream devices while maintaining accuracy. We're committed to finding the right balance between speed and accuracy.
Comparison: Time to Process 10 Minutes of Audio
| Device | Whisper Standard | Whisper Large-v3 Turbo | Improvement |
|---|---|---|---|
| iPhone 15 Pro | 425 seconds | 82 seconds | 5.2× faster |
| iPad Pro M2 | 380 seconds | 71 seconds | 5.4× faster |
| MacBook Pro M2 | 316 seconds | 63 seconds | 5.0× faster |
10 minutes of audio now transcribes in about a minute on recent hardware. Same accuracy, way faster.
Mac Version
Mac Whisper Notes uses the Turbo model on Apple Silicon Macs. M-series chips handle the faster processing really well—transcription speed matches the numbers above.
Download for iOS / macOS