Back to Overview

On-Device Transcription

Download Core ML optimized Model

Use Sequence's local Core ML model powered by Parakeet V3 for for private speech-to-text transcription.

How Do Model Downloads Work?

Once you activate your license, you can download our optimized Core ML Parakeet V3 model - learn more about that here, we host the models securely and can only be accessed via your account in app.

  1. Open Settings and choose Transcription.
  2. Tap the download button and the model is then stored on your device (this file is ~1GB)
  3. When you run your first transcription, the model can take some time to unpack and cache the model so subsequent runs are a lot faster.

Note: When you finish a transcription, we unload the model from your system memory which keeps Sequence at a really minimal footprint, when you trigger another transcription, it loads up again - this does mean if you wait long periods before running transcriptions you will loose a bit of speed, compared to running lots back to back whilst the model is hot and ready.