← All work

project - AI Systems

Music & Lyrics Transcription System

An AI-driven pipeline for extracting lyrical and melodic content from uploaded audio.

Role
Developer
Team / Org
Team project
Timeline
Oct 2022 – Nov 2022
Skills / Tools
  • Audio ML
  • ASR
  • Evaluation
  • Frontend clarity

Overview

I built an end-to-end transcription system that combined SpeechBrain and Spleeter to process uploaded tracks, evaluate reliability under different conditions, and present outputs through a React-based interface designed for clarity and interpretability.

What I did

  • Built an end-to-end audio processing pipeline for lyrics and melody extraction.

  • Benchmarked transcription performance using WER and F1.

  • Identified reliability degradation under high-tempo conditions.

Outcomes

0.75
F1
18–20%
WER