Conventional automatic speech recognition (ASR) often neglects the spectral phase information in its front-end and feature extraction stages. The aim of this paper is to show the impact that enhancement of the noisy spectral phase has on ASR accuracy when dealing with speech signals corrupted with additive noise. Apart from proof-of-concept experiments using clean spectral phase, we also present a phase enhancement method as a phase-aware front-end and modified group delay as a phase-aware feature extractor, and the combination thereof. In experiments, we demonstrate the improved performance for each individual component and their combination, compared to the conventional phase-unaware Mel Frequency Cepstral Coefficients (MFCCs)-based ASR. We observe that the estimated phase information used in the front-end or feature extraction component improves the ASR word accuracy rate (WAR) by 20.98% absolute for noise corrupted speech (averaged over SNRs ranging from 0 to 20 dB).
Cite as: Fahringer, J., Schrank, T., Stahl, J., Mowlaee, P., Pernkopf, F. (2016) Phase-Aware Signal Processing for Automatic Speech Recognition. Proc. Interspeech 2016, 3374-3378, doi: 10.21437/Interspeech.2016-823
@inproceedings{fahringer16_interspeech,
author={Johannes Fahringer and Tobias Schrank and Johannes Stahl and Pejman Mowlaee and Franz Pernkopf},
title={{Phase-Aware Signal Processing for Automatic Speech Recognition}},
year=2016,
booktitle={Proc. Interspeech 2016},
pages={3374--3378},
doi={10.21437/Interspeech.2016-823}
}