PiCoGen: Generate Piano Covers with a Two-stage Approach

TL;DR

PiCoGen is a two-stage system that automatically generates piano covers from audio recordings by transcribing melodies and chords, then synthesizing a piano performance without needing paired training data.

Contribution

It introduces a novel two-stage approach for cover song generation that does not require paired datasets, outperforming existing methods across genres.

Findings

PiCoGen achieves competitive or superior performance compared to existing methods.

The approach works across different musical genres.

It effectively transcribes melodies and chords from audio recordings.

Abstract

Cover song generation stands out as a popular way of music making in the music-creative community. In this study, we introduce Piano Cover Generation (PiCoGen), a two-stage approach for automatic cover song generation that transcribes the melody line and chord progression of a song given its audio recording, and then uses the resulting lead sheet as the condition to generate a piano cover in the symbolic domain. This approach is advantageous in that it does not required paired data of covers and their original songs for training. Compared to an existing approach that demands such paired data, our evaluation shows that PiCoGen demonstrates competitive or even superior performance across songs of different musical genres.