New Missing Features Mask Estimation Method for Speaker Recognition in Noisy Environments

摘要

Currently, many speaker recognition applications must handle speech corrupted by environmental additive noise without having a priori knowledge about the characteristics of noise. Some previous works in speaker recognition have used Missing Feature (MF) approach to compensate for noise. In most of those applications the spectral reliability decision step is done using the Signal to Noise Ratio (SNR) criterion. This has the goal of enhancing signal power rather than noise power, which could be dangerous in speaker recognition tasks, because useful speaker information could be removed. This work proposes a new mask estimation method based on Speaker Discriminative Information (SDI) for determining spectral reliability in speaker recognition applications based on the MF approach. The proposal was evaluated through speaker verification experiments in speech corrupted by additive noise. Experiments demonstrated that this new criterion has a promising performance in speaker verification tasks.