A real-time speech recognition system based on the Implementation of FPGA

The application of embedded speech recognition system was mainly implemented on human computer interaction, interactive voice response system, military etc. The design scheme of embedded speech recognition system was implemented to base on the Cyclone II-EP2C35F672C6 of field-programmable gate array (FPGA) and the soft-core processor of NIOS-II. A real-time speech recognition engine which combined the implementation of hardware with the implementation of software was implemented to take as an input the time domain signal from a microphone and perform the frequency domain feature extraction on the sample to identify the word being spoken on the FPGA. It was about 17 times faster than the implementation of C program. A higher ratio of speech recognition was reached by the method of developing software and hardware on the FPGA.