Influence of Noisy Environment on the Speech Recognition Rate Based on the Altera FPGA

Abstract

This paper introduce an approach to study the effects of different levels ofenvironment noise on the recognition rate of speech recognition systems, which arenot used any type of filters to deal with this issue. This is achieved by implementingan embedded SoPC (System on a Programmable Chip) technique with Altera Nios IIprocessor for real-time speech recognition system. Mel Frequency CepstralCoefficients (MFCCs) technique was used for speech signal feature extraction(observation vector). Model the observation vector of voice information by usingGaussian Mixture Model (GMM), this model passed to the Hidden Markov Model(HMM) as probabilistic model to process the GMM statistically to make decision onutterance words recognition, whether a single or composite, one or more syllablewords. The framework was implemented on Altera Cyclone II EP2C70F896C6NFPGA chip sitting on ALTERA DE2-70 Development Board. Each word model(template) stored as Transition Matrix, Diagonal Covariance Matrices, and MeanVectors in the system memory. Each word model utilizes only 4.45Kbytes regardlessof the spoken word length. Recognition words rate (digit/0 to digit/10) given 100%for the individual speaker. The test was conducted at different sound levels of thesurrounding environment (53dB to 73dB) as measured by Sound Level Meter (SLM)instrument.