Normalized amplitude modulation features for large vocabulary noise-robust speech recognition

Abstract

Background noise and channel degradations seriously constrain the performance of state-of-the-art speech recognition systems. Studies comparing human speech recognition performance with automatic speech recognition systems indicate that the human auditory system is highly robust against background noise and channel variabilities compared to automated systems. A traditional way to add robustness to a speech recognition system is to construct a robust feature set for the speech recognition model. In this work, we present an amplitude modulation feature derived from Teager's nonlinear energy operator that is power normalized and cosine transformed to produce normalized modulation cepstral coefficient (NMCC) features. The proposed NMCC features are compared with respect to state-of-the-art noise-robust features in Aurora-2 and a renoised Wall Street Journal (WSJ) corpus. The WSJ word-recognition experiments were performed on both a clean and artificially renoised WSJ corpus using SRI's DECIPHER large vocabulary speech recognition system. The experiments were performed under three train-test conditions: (a) matched, (b) mismatched, and (c) multi-conditioned. The Aurora-2 digit recognition task was performed using the standard HTK recognizer distributed with Aurora-2. Our results indicate that the proposed NMCC features demonstrated noise robustness in almost all the training-test conditions of renoised WSJ data and also improved digit recognition accuracies for Aurora-2 compared to the MFCCs and state-of-the-art noise-robust features

Auditory processing-based features for improving speech recognition in adverse acoustic conditions

“…For improving robustness, the normalized modulation spectra have been proposed in [23]. Similar work in the context of large vocabulary speech recognition such as noisy Wall Street Journal (New York, NY, USA) and GALE task as reported in [24,25].…”

Calibration and multiple system fusion for spoken term detection using linear logistic regression

“…The MFCCs were also augmented with a 10-dimensional voicing feature vector [12]. The three novel features explored were: (1) The Normalized Modulation Cepstral Coefficient (NMCC) [13], obtained from tracking the amplitude modulations of the sub-band speech signals in time domain. The produced 52-dimensional vector was reduced to 20 with principal component analysis (PCA) (NMCC20).…”

Medium-duration modulation cepstral feature for robust speech recognition

“…followed by cepstral feature extraction; or (2) by using noise robust speech-processing approaches, where noiserobust transforms and/or human perception based speech analysis methodologies are deployed for acoustic-feature generation (e.g., ETSI [European Telecomm. Standards Institute] advanced frontend [4], power normalized cepstral coefficients [PNCC] [5], modulation based features [6,7], and several others).…”