An Implementation of Viterbi Algorithm on GPU

General purpose computation based on GPU is a hot topic for research in recent years. The paper presents the parallel implementation of Viterbi algorithm on GPU based on features of GPU and characteristics of Viterbi algorithm in keyword spotting system. The results of examination by using NVIDIA 9600 GT GPU show that the GPU, in comparison to traditional processing platform, could enhance the processing performance if the recognition accuracy of keyword spotting system is ensured.

[1]  P. J. Narayanan,et al.  High Performance Pattern Recognition on GPU , 2008 .

[2]  Andrew J. Viterbi,et al.  Error bounds for convolutional codes and an asymptotically optimum decoding algorithm , 1967, IEEE Trans. Inf. Theory.

[3]  Sadaoki Furui,et al.  The Titech large vocabulary WFST speech recognition system , 2007, 2007 IEEE Workshop on Automatic Speech Recognition & Understanding (ASRU).

[4]  Douglas D. O'Shaughnessy,et al.  Recognition of digit strings in noisy speech with limited resources , 2000, INTERSPEECH.

[5]  Eric A. Brewer,et al.  Hardware speech recognition for user interfaces in low cost, low power devices , 2005, Proceedings. 42nd Design Automation Conference, 2005..

[6]  Rob A. Rutenbar,et al.  Moving speech recognition from software to silicon: the in silico vox project , 2006, INTERSPEECH.

[7]  Meichun Hsu,et al.  GPU-Accelerated Large Scale Analytics , 2009 .

[8]  Kurt Keutzer,et al.  Data-Parallel Large Vocabulary Continuous Speech Recognition on Graphics Processors , 2008 .

[9]  J. Kulpa,et al.  Time-frequency analysis using NVIDIA compute unified device architecture (CUDA) , 2009, Symposium on Photonics Applications in Astronomy, Communications, Industry, and High-Energy Physics Experiments (WILGA).

[10]  Jong Kyoung Kim,et al.  Speech recognition , 1983, 1983 IEEE International Solid-State Circuits Conference. Digest of Technical Papers.

[11]  Wayne H. Ward,et al.  Speech recognition , 1997 .