Singing Voice Audio Dataset

This dataset is for the purpose of the analysis of singing voice. It is our hope that the publication of this dataset will encourage further work into the area of singing voice audio analysis by removing one of the main impediments in this research area - the lack of data (unaccompanied singing).

It contains over 70 original vocal recordings by 28 professional, semi-professional and amateur singers. Singing style is predominantly Chinese Opera but some recordings are Western Opera. All recordings are 44.1 KHz sample rate and have been amplitude normalised.

It comes with an Excel spreadsheet describing each audio file. This is included in the dataset and available as a separate download. Whilst every effort has been made to ensure the accuracy of the song information, we apologise for any errors and ask that you contact if you discover any mistakes.

This is the first release of the dataset and only unaccompanied audio has been included. Whilst the spread-sheet indicates some recordings are also available with musical accompaniment, these are not included in this release as they require mixing.

It is our intention to continually add to this database as more recordings are made and to add feature sets and annotations as they become available.

The dataset can be downloaded using the following link:

SingingDatabase (zip, size: ~1.5GB)

metadata descriptors (xlsx)

These files are also available through the C4DM Research Data Repository here.

These audio samples are released under the Creative Commons Attribution-Noncommercial-Share-Alike license with attribution to the Centre for Digital Music, Queen Mary, University of London.

If you use this database please cite : Black, Dawn A. A., Li, Ma and Tian, Mi. "Automatic Identification of Emotional Cues in Chinese Opera Singing", in Proc. of 13th Int. Conf. on Music Perception and Cognition and the 5th Conference for the Asian-Pacific Society for Cognitive Sciences of Music (ICMPC 13-APSC0M 5 2014), Seoul, South Korea, August 2014.