A Neural Network based local SNR estimation for estimating spectral masks

In this work, we present a new mask estimation technique that uses a neural network classifier to determine the reliability of spectrographic elements. In addition some different kinds of features used for classification were compared that make no assumptions about the corrupting noise signal, but rather exploit spectrographic characteristics of the speech signal. The performance of the proposed method is experimentally evaluated in text independent speaker recognition task using the Gaussian mixture model (GMM) under various noise conditions. The speaker recognition results were achieved using the TFarsdat corpus. Noisy speech is simulated by adding noise sources taken from the NOISEX-92 database. Experimental results obtained show that the new neural network based mask estimation method is effective for speaker recognition under noisy conditions.