Unsupervised Equalization of Lombard Effect for Speech Recognition in Noisy Adverse Environments

• Published in: IEEE transactions on audio, speech, and language processing Vol. 18; no. 6; pp. 1379 - 1393
• Main Authors: Boril, Hynek, Hansen, John H L
• Format: Journal Article
• Language: English
• Published: Piscataway, NJ IEEE 01-08-2010; Institute of Electrical and Electronics Engineers
• Subjects: Acoustic noise; Applied sciences; Automatic speech recognition; Background noise; Cepstral analysis; Cepstral compensation; codebook of noisy models; Coding, codes; Detection, estimation, filtering, equalization, prediction; Exact sciences and technology; Frequency domain analysis; frequency warping; Information, signal and communications theory; Lombard effect; Loudspeakers; Noise reduction; Signal and communications theory; Signal processing; Signal representation. Spectral analysis; Signal, noise; Speech enhancement; Speech processing; Speech recognition; Telecommunications and information theory; Working environment noise; Performance evaluation; Error rate; Unsupervised classification; Background noise; Absolute error; Accuracy; Codebook; Model matching; Cepstral compensation; frequency warping; Background; Hostile environment; Frequency domain method; Decoding; Equalization; Algorithm; Signal classification; Mismatching; Cepstral analysis; Time domain method; Ambient noise; Speech recognition; codebook of noisy models; Maximum likelihood; Automatic recognition; Lombard effect; Signal to noise ratio
• ISSN: 1558-7916; 1558-7924
• DOI: 10.1109/TASL.2009.2034770

In the presence of environmental noise, speakers tend to adjust their speech production in an effort to preserve intelligible communication. The noise-induced speech adjustments, called Lombard effect (LE), are known to severely impact the accuracy of automatic speech recognition (ASR) systems. The reduced performance results from the mismatch between the ASR acoustic models trained typically on noise-clean neutral (modal) speech and the actual parameters of noisy LE speech. In this study, novel unsupervised frequency domain and cepstral domain equalizations that increase ASR resistance to LE are proposed and incorporated in a recognition scheme employing a codebook of noisy acoustic models. In the frequency domain, short-time speech spectra are transformed towards neutral ASR acoustic models in a maximum-likelihood fashion. Simultaneously, dynamics of cepstral samples are determined from the quantile estimates and normalized to a constant range. A codebook decoding strategy is applied to determine the noisy models best matching the actual mixture of speech and noisy background. The proposed algorithms are evaluated side by side with conventional compensation schemes on connected Czech digits presented in various levels of background car noise. The resulting system provides an absolute word error rate (WER) reduction on 10-dB signal-to-noise ratio data of 8.7% and 37.7% for female neutral and LE speech, respectively, and of 8.7% and 32.8% for male neutral and LE speech, respectively, when compared to the baseline recognizer employing perceptual linear prediction (PLP) coefficients and cepstral mean and variance normalization.