J.M. de Veth

Comparison of channel normalisation techniques for automatic speech recognition over the phone

We compared three different channel normalisation (CN) methods in the context of a connected digit recognition task over the phone: cepstrum mean substraction (CMS), RASTA filtering and the Gaussian dynamic cepstrum representation (GDCR). Using a small set of context independent (CI) continuous Gaussian mixture hidden Markov models (HMMs), we found that CMS and RASTA outperformed the GDCR technique. We show that the main cause for the superiority of CMS compared to RASTA is the phase distortion introduced by the RASTA filter. Recognition results for a phase corrected RASTA technique are identical to those of CMS. Our results indicate that an ideal cepstrum based CN method should: (1) effectively remove the DC component; (2) at least preserve modulation frequencies in the range 2-16 Hz; and (3) introduce no phase distortion in case CI HMMs are used for recognition.

Comparing acoustic features for robust ASR in fixed and cellular network applications

Within the context of automatic speech recognition (ASR) applications for telephony, we investigate the acoustic preprocessing issues that are at stake in going from the fixed line to the cellular network. Because the spectral representation used in enhanced full rate GSM is linear prediction, we investigate the relative advantages and drawbacks of conventional mel-frequency cepstral coefficient (MFCC) parameters derived from a non-parametric fast Fourier transform (FFT) and MFCC parameters derived from a linear predictive coding (LPC) spectral estimate. Robust formant parameters, also derived from an LPC description of the spectrum, are studied as an alternative to MFCCs. Within the framework of connected digit recognition based on hidden Markov models, ASR performance was measured for clean conditions, as well as for three different additive noise conditions. In addition, the performance of a conventional recognition procedure was compared with the performance of an ASR system based on our acoustic backing-off implementation of missing feature theory (MFT).

The impact of spectral and energy mismatch on the Aurora2 digit recognition task

Within the Aurora2 experimental framework, the aim of this study is to determine what the relative contributions of spectral shape and energy features are to the mismatch observed between clean training and noisy test data. In addition to measurements on the baseline Aurora2 system, recognition performance was also evaluated after the application of time domain noise reduction (TDNR) and histogram normalisation (HN) in the cepstral domain. The results indicate that, for the Aurora2 digit recognition task, TDNR, HN, as well as a combination of the two techniques achieve higher recognition rates by reducing the mismatch in the energy part of acoustic feature space. The corresponding mismatch reduction in the spectral shape features yields hardly any gain in recognition performance.