Bayesian neural network chip design for speech recognition system

Jhing Fa Wang; An Nan Suen; Jia Ru Lee; Chung Hsien Wu

Bayesian neural network chip design for speech recognition system

Jhing Fa Wang, An Nan Suen, Jia Ru Lee, Chung Hsien Wu

資訊工程學系

研究成果: Paper › 同行評審

2 引文斯高帕斯（Scopus）

摘要

The Bayesian Neural Network (BNN) has been widely used as speech recognition template which combines the merits of the Dynamic Programming (DP) and Hidden Markov Model (HMM) methods. However, it is computationally intensive and very costly to implement using DSP component. A single chip implementation of the BNN will drastically reduce the cost and the size of many speech recognition systems. It will also make low cost implementation of real-time speech recognition system possible. In this paper, the implementation of single BNN chip for the real-time speech recognizer is presented. Fabricated in 0.8 μm double-metal CMOS technology, the chip contains approximately 13000 transistors which occupy a 3.1×3.2 mm² area and has been tested to be fully functional at IMS XL-60 tester.

原文	English
頁面	2027-2030
頁數	4
出版狀態	Published - 1995 十二月 1
事件	Proceedings of the 1995 IEEE International Conference on Neural Networks. Part 1 (of 6) - Perth, Aust 持續時間: 1995 十一月 27 → 1995 十二月 1

Other

Other	Proceedings of the 1995 IEEE International Conference on Neural Networks. Part 1 (of 6)
城市	Perth, Aust
期間	95-11-27 → 95-12-01

All Science Journal Classification (ASJC) codes

Software

指紋深入研究「Bayesian neural network chip design for speech recognition system」主題。共同形成了獨特的指紋。

引用此

@conference{5168b824b10a40f781577e824f0ad1e4,

title = "Bayesian neural network chip design for speech recognition system",

abstract = "The Bayesian Neural Network (BNN) has been widely used as speech recognition template which combines the merits of the Dynamic Programming (DP) and Hidden Markov Model (HMM) methods. However, it is computationally intensive and very costly to implement using DSP component. A single chip implementation of the BNN will drastically reduce the cost and the size of many speech recognition systems. It will also make low cost implementation of real-time speech recognition system possible. In this paper, the implementation of single BNN chip for the real-time speech recognizer is presented. Fabricated in 0.8 μm double-metal CMOS technology, the chip contains approximately 13000 transistors which occupy a 3.1×3.2 mm2 area and has been tested to be fully functional at IMS XL-60 tester.",

author = "Wang, {Jhing Fa} and Suen, {An Nan} and Lee, {Jia Ru} and Wu, {Chung Hsien}",

year = "1995",

month = dec,

day = "1",

language = "English",

pages = "2027--2030",

note = "Proceedings of the 1995 IEEE International Conference on Neural Networks. Part 1 (of 6) ; Conference date: 27-11-1995 Through 01-12-1995",

}

TY - CONF

T1 - Bayesian neural network chip design for speech recognition system

AU - Wang, Jhing Fa

AU - Suen, An Nan

AU - Lee, Jia Ru

AU - Wu, Chung Hsien

PY - 1995/12/1

Y1 - 1995/12/1

N2 - The Bayesian Neural Network (BNN) has been widely used as speech recognition template which combines the merits of the Dynamic Programming (DP) and Hidden Markov Model (HMM) methods. However, it is computationally intensive and very costly to implement using DSP component. A single chip implementation of the BNN will drastically reduce the cost and the size of many speech recognition systems. It will also make low cost implementation of real-time speech recognition system possible. In this paper, the implementation of single BNN chip for the real-time speech recognizer is presented. Fabricated in 0.8 μm double-metal CMOS technology, the chip contains approximately 13000 transistors which occupy a 3.1×3.2 mm2 area and has been tested to be fully functional at IMS XL-60 tester.

AB - The Bayesian Neural Network (BNN) has been widely used as speech recognition template which combines the merits of the Dynamic Programming (DP) and Hidden Markov Model (HMM) methods. However, it is computationally intensive and very costly to implement using DSP component. A single chip implementation of the BNN will drastically reduce the cost and the size of many speech recognition systems. It will also make low cost implementation of real-time speech recognition system possible. In this paper, the implementation of single BNN chip for the real-time speech recognizer is presented. Fabricated in 0.8 μm double-metal CMOS technology, the chip contains approximately 13000 transistors which occupy a 3.1×3.2 mm2 area and has been tested to be fully functional at IMS XL-60 tester.

UR - http://www.scopus.com/inward/record.url?scp=0029514615&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0029514615&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:0029514615

SP - 2027

EP - 2030

T2 - Proceedings of the 1995 IEEE International Conference on Neural Networks. Part 1 (of 6)

Y2 - 27 November 1995 through 1 December 1995

ER -