Versebatim

Project Overview

Versebatim turns real music into language lessons. Every song in the catalog exists in each supported language, every lyric line is synced to the beat, and one tap flips the whole track into the language you're learning — same melody, same beat, resuming on the exact line, so you hear the meaning land in real time. Behind the player sits a full content pipeline: songs are transcribed with OpenAI Whisper for word-level timestamps, the written lyrics are aligned to the audio by a custom dynamic-programming sequence aligner built to survive repeated choruses and verses, and translations are reviewed and fine-tuned against live playback in an internal studio tool before anything is published. The product ships as a mobile app and a web player sharing one TypeScript monorepo backed by Supabase.

Technologies Used

Key Features

• The signature flip — one tap swaps the whole song to another language, resuming on the exact line
• Synced lyrics that highlight and auto-scroll with the track, so your eyes never lose the singer
• Six languages across the catalog — English, Spanish, Portuguese, Italian, German, and Chinese
• Custom song pipeline: Whisper transcription plus a dynamic-programming lyric aligner produce word-level timestamps for every track
• Internal studio tool for reviewing timestamps, tuning them against playback, and managing translations
• Expo / React Native mobile app and web player in one TypeScript monorepo, backed by Supabase