Using ASR posteriors instead of UBM posteriors to train i-vectors. Based on the idea proposed by Yun Lei et al., "A novel scheme for speaker recognition using a phonetically-aware deep neural network," ICASSP14
Shell 0.62%Scilab 94.97%MATLAB 2.04%Python 2.37%
supervised_ivectors's Introduction
This project is build the recently popularized idea of training ivectors using ASR posteriors.
Original paper:
Yun Lei; Scheffer, N.; Ferrer, L.; McLaren, M., "A novel scheme for speaker recognition using a phonetically-aware deep neural network," in Acoustics, Speech and Signal Processing (ICASSP), 2014 IEEE International Conference on , vol., no., pp.1695-1699, 4-9 May 2014
The main project is on the CRSS cluster. But the original repo was built on my local machine.
Required packages:
MSR_identity_toolkit --> For MATLAB experiments.
Navid Shokouhi
September 2015