3 ms·

Why can't they just have full on-device transcription models? It's lossy, but extremely compressed.

• jankiehodgpodge 6y ago
  Because they're cheap, low powered devices.
  • MiroF 6y ago
    Even cheap devices can run transcription inference.