Efficient codebooks for fast and accurate low resource ASR systems

• Published in: Speech communication Vol. 51; no. 9; pp. 732 - 743
• Main Authors: Zouari, Leila, Chollet, Gérard
• Format: Journal Article Conference Proceeding
• Language: English
• Published: Amsterdam Elsevier B.V 01.09.2009; Elsevier; Elsevier : North-Holland
• Subjects: Applied sciences; Codebook; Computer Science; Exact sciences and technology; Gaussian selection; Information, signal and communications theory; Signal processing; Speech processing; Speech recognition; Telecommunications and information theory; Codebook; Speech recognition; Gaussian selection; Performance evaluation; Cost minimization; Probabilistic approach; Man machine dialogue; Speech-based user interfaces; Algorithm; Hidden Markov models; Metric; Automatic recognition; Multilevel system; Speech processing; codebook
• ISSN: 0167-6393; 1872-7182
• DOI: 10.1016/j.specom.2009.01.010

Today, speech interfaces have become widely employed in mobile devices, thus recognition speed and resource consumption are becoming new metrics of Automatic Speech Recognition (ASR) performance. For ASR systems using continuous Hidden Markov Models (HMMs), the computation of the state likelihood is one of the most time consuming parts. In this paper, we propose novel multi-level Gaussian selection techniques to reduce the cost of state likelihood computation. These methods are based on original and efficient codebooks. The proposed algorithms are evaluated within the framework of a large vocabulary continuous speech recognition task.