In part 1 I went over a basic approach of how web browsers support recording audio and my approach to using an on-device Voice to Text model, specifically OpenAI’s Whisper optimized for the edge with transformers.js.