Feature Warping for Robust Speaker Verification

TLDR

Adverse channel mismatch, noise, and handset transducer effects distort the short‑term distribution of speech features, and existing methods only partially align these statistics to a target distribution. The study proposes a robust feature mapping approach to mitigate channel mismatch, additive noise, and handset transducer non‑linear effects. The method warps cepstral feature distributions to a standardized target over a specified interval and is evaluated against Gaussian target mapping and other enhancement techniques for speaker verification. The warping technique outperforms methods such as CMS, modulation spectrum processing, and short‑term windowed CMS and variance normalisation in speaker verification.

Abstract

We propose a novel feature mapping approach that is robust to channel mismatch, additive noise and to some extent, non-linear effects attributed to handset transducers. These adverse effects can distort the short-term distribution of the speech features. Some methods have addressed this issue by conditioning the variance of the distribution, but not to the extent of conforming the speech statistics to a target distribution. The proposed target mapping method warps the distribution of a cepstral feature stream to a standardised distribution over a specified time interval. We evaluate a number of the enhancement methods for speaker verification, and compare them against a Gaussian target mapping implementation. Results indicate improvements of the warping technique over a number of methods such as Cepstral Mean Subtraction (CMS), modulation spectrum processing, and short-term windowed CMS and variance normalisation.