Download the Optimised Core ML Model

How model downloads work

After you activate a trial or licence, download Sequence’s encrypted Core ML conversion of NVIDIA’s Parakeet v3. Read why Sequence uses Core ML for the engineering background.

The model is hosted securely and can only be downloaded through your account in the app.

Open Settings → Transcription. If you are signed in with valid access, the model area shows the download action. The file is roughly 1 GB; download time depends on your connection.

Local transcription

After the model is installed, transcription runs locally on your Mac. Your media is not uploaded to a cloud transcription service. Sequence uses the downloaded model on-device, then stores the resulting transcript in your local encrypted project database.

First run and warm runs

The first transcription can take longer because Sequence prepares the model for your system. Later transcriptions benefit from that cached preparation.

When a transcription finishes, Sequence unloads the model from system memory to keep the app’s footprint low. If you run several transcriptions close together the model can feel quicker because the model is still hot in memory.

Managing the download

The Transcription settings page also shows the resolved model path and download state. If a download fails, you can try again from the same screen. If you no longer need the model, you can delete the downloaded copy and re-download it later when required.