Choose the local speech model that turns recordings into text.
The transcription settings make the speech engine explicit. Talkie can keep recognition local, with the active model visible before you record.