JaCappella Corpus: A Japanese a Cappella Vocal Ensemble Corpus

We construct a corpus of Japanese a cappella vocal ensembles (jaCappella corpus) for vocal ensemble separation and synthesis. It consists of 35 copyright-cleared vocal ensemble songs and their audio recordings of individual voice parts. These songs were arranged from out-of-copyright Japanese children's songs and have six voice parts (lead vocal, soprano, alto, tenor, bass, and vocal percussion). They are divided into seven subsets, each of which features typical characteristics of a music genre such as jazz and enka. The variety in genre and voice part match vocal ensembles recently widespread in social media services such as YouTube, although the main targets of conventional vocal ensemble datasets are choral singing made up of soprano, alto, tenor, and bass. Experimental evaluation demonstrates that our corpus is a challenging resource for vocal ensemble separation. Our corpus is available on our project page (https://tomohikonakamura.github.io/jaCappella_corpus/).

PDF Abstract

Datasets


Introduced in the Paper:

jaCappella
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Vocal ensemble separation jaCappella MRDLA SDR (vocals) 8.7 # 2
SDR (soprano) 11.8 # 1
SDR (alto) 14.7 # 1
SDR (tenor) 11.3 # 2
SDR (bass) 10.2 # 2
SDR (vocal percussion) 22.1 # 1
Vocal ensemble separation jaCappella DPTNet SDR (vocals) 8.9 # 1
SDR (soprano) 8.5 # 3
SDR (alto) 11.9 # 3
SDR (tenor) 14.9 # 1
SDR (bass) 19.7 # 1
SDR (vocal percussion) 21.9 # 2
Vocal ensemble separation jaCappella X-UMX SDR (vocals) 7.5 # 3
SDR (soprano) 10.7 # 2
SDR (alto) 13.5 # 2
SDR (tenor) 10.2 # 3
SDR (bass) 9.1 # 3
SDR (vocal percussion) 21 # 3

Methods


No methods listed for this paper. Add relevant methods here