Music Product

TABLURA

A browser-based guitar learning app built around real songs: synced tab, AI transcription, isolated stems, and note-by-note practice feedback.

Tablura icon placeholder

Placeholder visual using the current app icon from the Tablura project.

Product Focus

Tablura is positioned as a practice product rather than a tab archive. The core loop is simple: bring in a song, follow the synced tab, and get instant feedback when you hit the right notes.

Technical Shape

The app combines browser audio processing, AI inference, notation rendering, account state, and paid feature gating in one product surface. That means frontend UX, compute-heavy client features, and product monetization all have to align.

Key Capabilities

  • Synced tab playback for imported songs and Guitar Pro files
  • AI transcription from uploaded audio files
  • Stem separation to isolate guitar before transcription
  • Real-time pitch detection and practice feedback
  • Supabase-backed authentication and user library
  • Stripe subscriptions for gated Plus features

Stack Highlights

Next.jsTypeScriptalphaTabTensorFlow.jsWebGPUBasic PitchSupabaseStripeZustand

The implementation leans on `alphaTab` for notation playback, Spotify Basic Pitch and TensorFlow for transcription, Supabase for auth and persistence, and Stripe for recurring billing.

Why It Belongs In This Portfolio

Tablura fits the same pattern as the rest of the recent work: product experience tied directly to operational systems. In this case the interesting boundary is music UX plus browser-side AI, user accounts, subscriptions, and the mechanics required to turn audio into a playable learning workflow.