Hardware Accelerated FPGA Implementation of Lithuanian Isolated Word Recognition System

摘要

Speech is the most natural way of human communication. At least half of a century researchers and engineers are trying to automate speech recognition and synthesis. Yet there is no automatic speech recognition system equivalent to human recognition with its precision, robustness to noise and speaker, and versatility. Despite of difficulties of implementation it is agreed that speech is the future interface between computer and the human. Although software implementations of speech analysis and synthesis systems are capable of real time operating, growing number of mobile devices implies the need for hardware implementation of speech processing systems. One technique of the potential speech recognition hardware implementations is Field Programmable Gate Array (FPGA). It is an array of semiconductor logical and memory elements interconnected by reprogrammable logical links. FPGA enables faster, less time consuming hardware implementation and finds a lot of applications in various fields: speech and signal processing [1-3], image processing [4], optimization [5] and networking [6].