An FPGA-based voice signal preprocessor for the real-time cross-correlation

摘要

In this paper, we present an FPGA-based real-time voice signal preprocessing system that can compute the cross-correlation of the voice signals from number of microphones. Each cross-correlation is calculated using voice signals captured within 200 ms which has 10 ms shifting window. The sampling frequency of the proposed system is 48 KHz and the resolution of each sample is 16 bits. The entire algorithm is implemented in hardware based on FPGA and runs in real-time.