Transcription feat: Detect the presence of a specific sound in the audio stream (is a person speaking?) https://developers.google.com/mediapipe/solutions/audio/audio_classifier/web_js perf: Externalize DSP processing into isolated library / (e.g. using WebAssembly)