Multiple Approaches to Robust Speech Recognition

MULTIPLE APPROACHES TO ROBUST SPEECH RECOGNITION Richard M. Stern, Fu-Hua Liu, Yoshiaki Ohshima, Thomas M. Sullivan, Alejandro Acero* Department of Electrical and Computer Engineering School of Computer Science Carnegie Mellon University Pittsburgh, PA 15213 ABSTRACT This paper compares several different approaches to robust speech recognition. We review CMU's ongoing research in the use of acoustical pre-proeessing to achieve robust speech recog- nition, and we present the results of the first evaluation of pre- processing in the context of the DARPA standard ATIS domain for spoken language systems. We also describe and compare the effectiveness of three complementary methods of signal process- ing for robust speech recognition: acoustical pre-procossing, microphone array processing, and the use of physiologically- motivated models of peripheral signal processing. Recognition error rates are presented using these three approaches in isolation and in combination with each other for the speaker-independent continuous alphanumeric census speech recognition task. 1. INTRODUCTION The need for speech recognition systems and spoken lan- guage systems to be robust with respect to their acoustical environment has become more widely appreciated in recent years (e.g. [1]). Results of several studies have demonstrated that even automatic speech recognition systems that are designed to be speaker independent can perform very poorly when they are tested using a different type of microphone or acous- tical environment from the one with which they were trained (e.g. [2, 3]), even in a relatively quiet office en- vironment. Applications such as speech recognition over telephones, in automobiles, on a factory floor, or outdoors demand an even greater degree of environmental robust- ness. The CMU speech group is committed to the development of speech recognition systems that are robust with respect to environmental variation, just as it has been an early proponent of speaker-independent recognition. While most of our work presented to date has described new acoustical pre-processing algorithms (e.g. [2, 4, 5], we have always regarded pre-processing as one of several approaches that must be developed in concert to achieve robust recog- nition. The purpose of this paper is twofold. First, we describe our results for the DARPA benchmark evaluation for robust speech recognition for the ATIS task, discussing the effectiveness of our methods of acoustical pre- preprocessing in the context of this task. Second, we describe and compare the effectiveness of three com- plementary methods of signal processing for robust speech recognition: acoustical pre-processing, microphone array processing, and the use of physiologically-motivated models of peripheral signal processing. 2. ACOUSTICAL PRE-PROCESSING We have found that two major factors degrading the per- formance of speech recognition systems using desktop microphones in normal office environments are additive noise and unknown linear filtering. We showed in [2] that simultaneous joint compensation for the effects of additive noise and linear filtering is needed to achieve maximal robustness with respect to acoustical differences between the training and testing environments of a speech recog- nition system. We described in [2] two algorithms that can perform such joint compensation, based on additive correc- tions to the cepstral coefficients of …

Content truncated to 500 words