Title :
Robust speech recognition using neural networks and hidden Markov models
Author :
Cong, Lin ; Asghar, Saf ; Cong, Bin
Author_Institution :
Adv. Micro Devices, Santa Clara, CA, USA
Abstract :
This paper proposes a robust, speaker-independent isolated word speech recognition (IWSR) system (SMQ/HMM-SVQ/HMM)/MLP which combines dual split matrix quantization (SMQ) and split vector quantization (SVQ) pair combined with both the strength of the HMM in modeling stochastic sequences and the non-linear classification capability of MLP neural networks (NN). The system efficiently utilizes processing resources and improves speech recognition performance by using neural networks as the classifier of the system. Computer simulation clearly indicates the superiority over conventional VQ/HMM and MQ/HMM systems with 98% and 95.8% recognition accuracy at 20 dB and 5 dB SNR levels, respectively in a car noise environment, based on the TIDIGIT database
Keywords :
acoustic noise; hidden Markov models; matrix algebra; multilayer perceptrons; quantisation (signal); speech recognition; stochastic processes; HMM; MLP neural networks; MQ/HMM system; SNR levels; TIDIGIT database; VQ/HMM system; car noise environment; computer simulation; dual split matrix quantization; hidden Markov models; multilayer perceptron; neural networks; nonlinear classification; processing resources; recognition accuracy; robust speech recognition; speaker-independent isolated word speech recognition; speech recognition performance; split vector quantization; stochastic sequences modeling; system classifier; Computer simulation; Hidden Markov models; Neural networks; Noise level; Robustness; Signal to noise ratio; Speech recognition; Stochastic systems; Vector quantization; Working environment noise;
Conference_Titel :
Information Technology: Coding and Computing, 2000. Proceedings. International Conference on
Conference_Location :
Las Vegas, NV
Print_ISBN :
0-7695-0540-6
DOI :
10.1109/ITCC.2000.844204
