Lesson scanner

live demo

Drop one of Nina's lesson recordings — acoustic fingerprint + the corpus metadata card, fully on-device

PRD F13 · Parallel track

Runs entirely on your device — the file is decoded and analyzed in the browser, never uploaded. This is the first working piece of the corpus pipeline from the founding meeting: 'AI code that can scan your videos… pull metadata… a per-video voice analyzer.' Drop any lesson recording (or any voice memo) to see its acoustic fingerprint.

🎞️

Drop a recording here

or tap to choose · audio works best (mp3/wav/m4a) · stays on your device

The full pipeline this seeds

Acoustic fingerprint (this page, real) → transcription + speaker diarization under a BAA (AssemblyAI ~$0.20/audio-hour, or self-hosted WhisperX) → lesson metadata like the card above → embeddings that ground the triage nurse in Nina's actual practice → Nina rating a few hundred clips (breathiness/strain/resonance) → the listening coach. On training data: the meeting's verdict stands — you want boththe lo-fi virtual recordings (they match real users' phone mics) and hi-fi ones, plus her written notes. Archive originals lossless; consent per student, always.