How to apply this guide
Check free disk space, download size and recovery before switching recognition models.
Start with
Choose the language and a supported smaller model first. Make a short recording that includes a name, a time and a last instruction. Upgrade only if the result misses what matters in your workflow.
Check before you finish
Record a fresh sample while disconnected after setup. Check that the app still lists the expected model and language. Delete an unused file only when you know how to download it again.
Evidence to keep: Keep the target field open and verify that the complete reviewed result reached the intended destination. This page’s key limit is: Check free disk space, download size and recovery before switching recognition models.
Start here
Local speech assets occupy storage and may need temporary room while downloading. Model sizes and hardware requirements differ across editions and devices.
Start with a compatible model
Choose the language and a supported smaller model first. Make a short recording that includes a name, a time and a last instruction. Upgrade only if the result misses what matters in your workflow.
Check before deleting
Choose a different installed model before removing the active one where possible. Use the app’s model management controls. Keep transcripts or retained recordings separate from model data.
Verify after changes
Record a fresh sample while disconnected after setup. Check that the app still lists the expected model and language. Delete an unused file only when you know how to download it again.




