FPGA Implementation of a Pipelined Gaussian Calculation for HMM-Based Large Vocabulary Speech Recognition. (3rd October 2010)

• Record Type: Journal Article
• Authors: Veitch, Richard; Aubert, Louis-Marie; Woods, Roger; Fischaber, Scott
• Other Names: Sutter Gustavo Academic Editor.
• Is Part Of: International journal of reconfigurable computing. Volume 2011(2011)
• Publication Date: 2010-10-03
• Subjects: Adaptive computing systems -- Periodicals; Adaptive computing systems; Periodicals; 004
• DOI: 10.1155/2011/697080
• Languages: English
• ISSNs: 1687-7195

Abstract

Abstract : A scalable large vocabulary, speaker independent speech recognition system is being developed using Hidden Markov Models (HMMs) for acoustic modeling and a Weighted Finite State Transducer (WFST) to compile sentence, word, and phoneme models. The system comprises a software backend search and an FPGA-based Gaussian calculation which are covered here. In this paper, we present an efficient pipelined design implemented both as an embedded peripheral and as a scalable, parallel hardware accelerator. Both architectures have been implemented on an Alpha Data XRC-5T1, reconfigurable computer housing a Virtex 5 SX95T FPGA. The core has been tested and is capable of calculating a full set of Gaussian results from 3825 acoustic models in 9.03 ms which coupled with a backend search of 5000 words has provided an accuracy of over 80%. Parallel implementations have been designed with up to 32 cores and have been successfully implemented with a clock frequency of 133 MHz.