Using Reconfigurable Computing in Speech Recognition Applications

Nowadays, during the past decades, reconfigurable computing (RC) has emerged as a technique for dynamically digital signal processing (DSP), especially machine recognition of speech. The typical machine recognition of speech has become into a signal processing by converting the speech waveform into a sequence of acoustic property vectors. It is possible by means of run-time reconfiguration computing systems, Field Programmable Gate Array (FPGA) s in particular based on Hidden Markov Models (HMMs). Due to the real time features of DSP, speech recognition applications need High Performance Computing (HPC) in RCs. In this paper, we highlighted the RC systems principles with focus on DSP application fields generally. Mainly, we discussed about machine recognition of speech by HMMs and modeling of speech processing and more study on using of the reconfigurable speech processing with introduction to prior works in this domain are considered. As a result, real-time speech recognition is a highly computationally demanding task can be possible in parallel processing systems by programmable RC. Keywords: Reconfigurable computing, Digital Signal Processing, Machine Learning, Speech Recognition;