Improving Audio Chord Estimation by Alignment and Integration of Crowd-Sourced Symbolic Music

Daphne Odekerken,Hendrik Vincent Koops,Anja Volk

doi:10.5334/tismir.81

Daphne Odekerken, Hendrik Vincent Koops + Show 1 more

Open Access

https://doi.org/10.5334/tismir.81

Copy DOI

Abstract

Automatic Chord Estimation (ACE) is a fundamental task in Music Information Retrieval (MIR) and has applications in both music performance and MIR research. The task consists of segmenting a music recording or score and assigning a chord label to each segment. Although it has been a task in the annual benchmarking evaluation MIREX for over 10 years, ACE is not yet a solved problem, since performance has stagnated and modern systems have started to tune themselves to subjective training data. We propose DECIBEL, a new ACE system that exploits heterogeneous musical representations, specifically MIDI and tab files, to improve audio-based ACE methods. From an audio file and a set of MIDI and tab files corresponding to the same popular music song, DECIBEL first estimates chord sequences. For audio, state-of-the-art audio ACE methods are used. MIDI files are aligned to the audio, followed by a MIDI chord estimation step. Tab files are transformed into untimed chord sequences and then aligned to the audio. Next, DECIBEL uses data fusion to integrate all estimated chord sequences into one final output sequence. DECIBEL improves all tested state-of-the-art ACE methods by 0.5 to 13.6 percentage points. This result shows that the integration of crowd-sourced annotations from heterogeneous symbolic music representations using data fusion is a suitable strategy for addressing challenging MIR tasks such as ACE.

Highlights

Automatic Chord Estimation (ACE) is an important task in Music Information Retrieval (MIR), with the goal of automatically estimating chords in audio recordings or symbolic music representations
It presents an opportunity to compare methods across different versions or to create methods that exploit the domain-specific strengths while attenuating their weaknesses in order to create higher quality analyses (e.g. Konz and Müller, 2012; Koops et al, 2016). In this light of cross-version analysis (CVA), we propose DECIBEL (DEtection of Chords Improved By Exploiting Linking symbolic formats), a novel system that exploits heterogeneous symbolic music representations, MIDI and tab files, for improving ACE on popular music
Using the existing one-to-one Music Information Retrieval Evaluation eXchange (MIREX) metrics, we find that DECIBEL improves each of the twelve tested ACE systems

Summary

Introduction

Automatic Chord Estimation (ACE) is an important task in Music Information Retrieval (MIR), with the goal of automatically estimating chords in audio recordings or symbolic music representations. ACE segments a musical piece so that the segment boundaries represent chord changes and each segment has a chord label. This is typically represented by a sequence of ⟨start time, end time, chord label⟩ triples. The estimation of chords in a musical piece is used in various MIR tasks, such as cover song identification, key detection, genre classification, lyrics interpretation and audio-to-lyrics alignment (McVicar et al, 2014). The main evaluation measure for ACE is (Weighted) Chord Symbol Recall (WCSR/CSR). State-of-theart ACE methods yield WCSRs of around 75–87%, given a chord vocabulary of major and minor chords.

Methods

Results

Conclusion