Music vocal analysis backend plan
This page lists the production backend systems needed to move this app from a browser prototype into a durable music vocal identity and AI voice reproducibility registry.
1. Account system
Add user accounts, login, artist ownership, team access, rights controller access, and secure access to each vocal identity record.
Recommended: Supabase Auth.
2. Permanent storage
Store guided vocal samples, artist vocal evidence files, registry records, consent records, reports, certificates, and exports on the backend instead of relying only on browser storage.
Current backend: Supabase database and Supabase Storage.
3. Server-side hashing
Generate official SHA-256 hashes on the backend after upload and store them in the audio metadata and export manifest.
Browser hashes can remain for instant feedback, but backend hashes should become the official record.
4. Music vocal analysis
Add real backend analysis for pitch contour, vocal range, vibrato rate and depth, formant and vowel profile, timbre profile, register behavior, breathiness, sibilance, onset and release behavior, melisma, slides, rhythm, phrasing, dynamics, and emotional delivery markers.
Recommended stack: Python, ffmpeg, librosa, praat-parselmouth, crepe or torchcrepe, matplotlib, and speaker-embedding models used only as supporting similarity reference.
5. AI reproducibility profile
Generate an AI voice reproducibility profile explaining which features make the singer recognizable and reproducible as a synthetic or cloned vocal model.
Focus areas: tone color, pitch behavior, vowel shape, phrasing, vibrato, runs, register changes, breath, pronunciation, and ad-lib style.
6. Report generation
Generate formal PDF reports that include hash tables, vocal range charts, waveform images, spectrogram images, pitch graphs, vibrato data, formant data, phrasing metrics, AI consent status, and export metadata.
7. Versioning
Store the guided vocal script version, browser analysis version, backend music-vocal analysis version, export format version, claim language version, and certificate version on every record.
8. Admin review status
Add an admin review workflow without claiming legal verification.
Use statuses like Submitted, Under Review, Accepted into Registry, Needs Correction, Withdrawn, and Archived.
This plan supports artist vocal identity documentation, AI voice consent-readiness, and synthetic voice protection. It does not claim legal ownership of a voice or make a legal finding of infringement.