Your personal AI scribe
Powered by Whisper Large v3 / Turbo
Live, unedited demo. Running Whisper Large v3 on a 3 year old iPhone 14 Pro.
Running on your phone, in real-time.
Truly Private
No servers, no cloud. ScribeAI runs on your device. Your data never leaves your phone.
Because everything runs on your phone the app works offline too.
State of the Art
Whisper Large v3 and Whisper Large v3 Turbo offer best in-class transcription accuracy but have always been too big to run directly on a phone.
Until now.
We have optimised both models so that for the first time they can run on an iPhone, in real time.
As our approach does not involve significant quantization you retain the full accuracy of the model. We do quantize to Q8, purely to make the model size on disk smaller. We find accuracy is maintained at this level compared to their native FP16. We can also run the models in real-time in FP16.
Simple pricing
Like the app? Just pay once and never pay again. No expiry, no subscription.
Subscription options are available too for people whom this is their prefered option. All subscriptions are offered with a 3 day free trial.
TranscribeKit
We are developing an SDK for developers who would like to use our tech in your own app. If you are interested please email us at hello@clinicalai.co.uk