L.H. Kim, S.I. Kang, H.S. Ryu, W.I. Jang, S.B. Lee, and A.S. Pandya
Speech recognition, DTW, VQ, wheelchair, fuzzy, DSP
In this study a speech recognition module has been developed and applied to a wheelchair for the physically handicapped. The main processor for the speech recognition module is TMS320C32, and we used Mel-Cepstrum 12 Order for the pre-processor step, which resulted in a substantial increase in the recognition rate in a noisy environment. DTW (Dynamic Time Warping) was used and proved to be excellent in performance for the speaker-dependent recognition part. In order to utilize this algorithm more effectively, the reference data were compressed to 1/12 using vector quantization so as to decrease the amount of memory required. In this article, a diverse range of technologies (endpoint detection, DMA processing, etc.) was employed in order to utilize the speech recognition system in real time. Furthermore, the overall wheelchair system configuration and its control algorithm were analyzed for performance.
Important Links:
Go Back
