Histogram Transform-based Speaker Identification

TL;DR

This paper introduces a new speaker identification method that employs histogram transform-based probability density functions of super-MFCC features, capturing dynamic speaker characteristics more effectively than traditional models.

Contribution

It proposes a novel histogram transform approach for PDF estimation in speaker ID, utilizing super-MFCC features to improve dynamic information capture.

Findings

HT model outperforms Gaussian mixture models in SI accuracy

Super-MFCC features enhance dynamic speaker characteristic representation

Histogram transform reduces histogram discontinuity issues

Abstract

A novel text-independent speaker identification (SI) method is proposed. This method uses the Mel-frequency Cepstral coefficients (MFCCs) and the dynamic information among adjacent frames as feature sets to capture speaker's characteristics. In order to utilize dynamic information, we design super-MFCCs features by cascading three neighboring MFCCs frames together. The probability density function (PDF) of these super-MFCCs features is estimated by the recently proposed histogram transform~(HT) method, which generates more training data by random transforms to realize the histogram PDF estimation and recedes the commonly occurred discontinuity problem in multivariate histograms computing. Compared to the conventional PDF estimation methods, such as Gaussian mixture models, the HT model shows promising improvement in the SI performance.