An ASIC based architecture for Vietnamese speech recognition on the basic of 130 nm TSMC technology

Abstract

Speech recognition has been researched over sixty years. The first researches were conducted from the 1950s till the early 1970s, some complete recognition systems were developed to recognise incoherent pronounciation in low noise conditions and only adapted to small word libraries (10-100 words) which belong to the trainers as well as the recognition persons. Today, the recognition systems adapting to the large word library are built based on computer system with high performance. Moreover, when integrated density is enhanced, the access to the hardware or handset applying recognition technology becomes feasible. In this work, an ASIC based architecture for Vietnamese speech recognition on the basis of 130 nm TSMC technology is illustrated to meet the real time requirements as well as confirm the highly effective performance.