The Implementation of Speech Recognition Systems on FPGA-Based Embedded Systems with SOC Architecture
Source: National University of Kaohsiung
An implementation of Artificial-Neural-Network (ANN) based speech recognition systems on the embedded platform is explored in this paper. An FPGA chip is adopted as the hardware of the embedded platform with the architecture of SOC. This makes the speech recognition systems applicable on the voice activated systems in toys, games, smart phones, office devices, vehicular communications, etc. In this paper, the authors use Integer FFT to replace the Floating FFT so that the speed of speech recognition is enhanced without influencing the recognition rate. The experimental results from the FPGA platform reveal that the speech recognition rate of the proposed hardware implementation methods is better than that in existing literatures.