Tether releases an open-weight speech model that transcribes audio offline
TranscribePsy keeps transcription on the device and offers dedicated phone builds. Tether reports strong speed and accuracy results, but the release waits until each recording or utterance ends.