Improved deep speaker feature learning for text-dependent speaker recognition

A deep learning approach has been proposed recently to derive speaker identifies (d-vector) by a deep neural network (DNN). This approach has been applied to text-dependent speaker recognition tasks and shows reasonable performance gains when combined with the conventional i-vector approach. Althoug...

Full description

Saved in:
Bibliographic Details
Published in2015 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA) pp. 426 - 429
Main Authors Lantian Li, Yiye Lin, Zhiyong Zhang, Dong Wang
Format Conference Proceeding
LanguageEnglish
Published Asia-Pacific Signal and Information Processing Association 01.12.2015
Subjects
d-vector
Data models
Machine learning
Mel frequency cepstral coefficient
Speaker recognition
time dynamic warping
Training
Training data
Online AccessGet full text
DOI10.1109/APSIPA.2015.7415306

Cover

More Information
Summary:A deep learning approach has been proposed recently to derive speaker identifies (d-vector) by a deep neural network (DNN). This approach has been applied to text-dependent speaker recognition tasks and shows reasonable performance gains when combined with the conventional i-vector approach. Although promising, the existing d-vector implementation still can not compete with the i-vector baseline. This paper presents two improvements for the deep learning approach: a phone-dependent DNN structure to normalize phone variation, and a new scoring approach based on dynamic time warping (DTW). Experiments on a text-dependent speaker recognition task demonstrated that the proposed methods can provide considerable performance improvement over the existing d-vector implementation.
DOI:10.1109/APSIPA.2015.7415306