Jingju Phoneme Classification Features for EUSIPCO 2017 paper

This dataset contains the pre-computed Mel-bands features from the training part of 

Rong Gong, Rafael Caro Repetto, & Yile Yang. (2017). Jingju a cappella singing dataset part1 [Data set]. Zenodo. http://doi.org/10.5281/zenodo.344932

This dataset is used for reproducing the singing voice phoneme classification experiment described in the following paper:

Timbre Analysis of Music Audio Signals with Convolutional Neural Networks  

For the usage and code, please refer to https://github.com/ronggong/EUSIPCO2017