Articulatory Information for Noise Robust Speech Recognition

• Published in: IEEE transactions on audio, speech, and language processing speech, and language processing, 2011-09, Vol.19 (7), p.1913-1924
• Main Authors: Mitra, V., Hosung Nam, Espy-Wilson, C. Y., Saltzman, E., Goldstein, L.
• Format: Article
• Language: English
• Subjects: Acoustics; Adaptation model; Applied sciences; Articulatory phonology; articulatory speech recognition; Artificial intelligence; artificial neural networks (ANNs); Computer science; control theory; systems; Connectionism. Neural networks; Exact sciences and technology; Feature recognition; Hidden Markov models; Information theory; Information, signal and communications theory; Mathematical models; Neural networks; Noise; Noise measurement; noise-robust speech recognition; Recognition; Signal processing; Speech; speech inversion; Speech processing; Speech recognition; task dynamic model; Tasks; Telecommunications and information theory; vocal-tract variables; Voice recognition
• ISSN: 1558-7916; 2329-9290; 1558-7924; 2329-9304
• DOI: 10.1109/TASL.2010.2103058

Prior research has shown that articulatory information, if extracted properly from the speech signal, can improve the performance of automatic speech recognition systems. However, such information is not readily available in the signal. The challenge posed by the estimation of articulatory information from speech acoustics has led to a new line of research known as "acoustic-to-articulatory inversion" or "speech-inversion." While most of the research in this area has focused on estimating articulatory information more accurately, few have explored ways to apply this information in speech recognition tasks. In this paper, we first estimated articulatory information in the form of vocal tract constriction variables (abbreviated as TVs) from the Aurora-2 speech corpus using a neural network based speech-inversion model. Word recognition tasks were then performed for both noisy and clean speech using articulatory information in conjunction with traditional acoustic features. Our results indicate that incorporating TVs can significantly improve word recognition rates when used in conjunction with traditional acoustic features.