Design a co-processor for output probability calculation in speech recognition

Li Peng; Hua Tang

doi:10.1109/ISCAS.2009.5117762

Design a co-processor for output probability calculation in speech recognition

Li Peng, Hua Tang

Electrical Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

2 Scopus citations

Abstract

In the CHMM (Continuous Hidden Markov Model) based speech recognition algorithm, Output Probability Calculation (OPC) is the most computation- intensive part. To reduce power consumption and design cost, this paper presents a custom-designed co-processor to implement OPC. The standard SRAM interface of the co-processor allows it to be controlled by various micro-controllers. The co-processor has been implemented in standard-cell based approach and manufactured in 0.18μm UMC technology. Tested with a 358-state 3-mixture 27-feature 800-word HMM, the co-processor operates at 10MHz to meet real-time requirement. The power consumption of this co-processor is 1.6mW, and the die size is 1.18mm².

Original language	English (US)
Title of host publication	2009 IEEE International Symposium on Circuits and Systems, ISCAS 2009
Pages	369-372
Number of pages	4
DOIs	https://doi.org/10.1109/ISCAS.2009.5117762
State	Published - Oct 26 2009
Event	2009 IEEE International Symposium on Circuits and Systems, ISCAS 2009 - Taipei, Taiwan, Province of China Duration: May 24 2009 → May 27 2009

Publication series

Name	Proceedings - IEEE International Symposium on Circuits and Systems
ISSN (Print)	0271-4310

Other

Other	2009 IEEE International Symposium on Circuits and Systems, ISCAS 2009
Country/Territory	Taiwan, Province of China
City	Taipei
Period	5/24/09 → 5/27/09

Access

10.1109/ISCAS.2009.5117762

OpenUrl availability

Full text

Cite this

Peng, L & Tang, H 2009, Design a co-processor for output probability calculation in speech recognition. in 2009 IEEE International Symposium on Circuits and Systems, ISCAS 2009., 5117762, Proceedings - IEEE International Symposium on Circuits and Systems, pp. 369-372, 2009 IEEE International Symposium on Circuits and Systems, ISCAS 2009, Taipei, Taiwan, Province of China, 5/24/09. https://doi.org/10.1109/ISCAS.2009.5117762

@inproceedings{faf5f819d05649b3a5ace9d6acb9e35c,

title = "Design a co-processor for output probability calculation in speech recognition",

abstract = "In the CHMM (Continuous Hidden Markov Model) based speech recognition algorithm, Output Probability Calculation (OPC) is the most computation- intensive part. To reduce power consumption and design cost, this paper presents a custom-designed co-processor to implement OPC. The standard SRAM interface of the co-processor allows it to be controlled by various micro-controllers. The co-processor has been implemented in standard-cell based approach and manufactured in 0.18μm UMC technology. Tested with a 358-state 3-mixture 27-feature 800-word HMM, the co-processor operates at 10MHz to meet real-time requirement. The power consumption of this co-processor is 1.6mW, and the die size is 1.18mm2.",

author = "Li Peng and Hua Tang",

year = "2009",

month = oct,

day = "26",

doi = "10.1109/ISCAS.2009.5117762",

language = "English (US)",

isbn = "9781424438280",

series = "Proceedings - IEEE International Symposium on Circuits and Systems",

pages = "369--372",

booktitle = "2009 IEEE International Symposium on Circuits and Systems, ISCAS 2009",

note = "2009 IEEE International Symposium on Circuits and Systems, ISCAS 2009 ; Conference date: 24-05-2009 Through 27-05-2009",

}

TY - GEN

T1 - Design a co-processor for output probability calculation in speech recognition

AU - Peng, Li

AU - Tang, Hua

PY - 2009/10/26

Y1 - 2009/10/26

N2 - In the CHMM (Continuous Hidden Markov Model) based speech recognition algorithm, Output Probability Calculation (OPC) is the most computation- intensive part. To reduce power consumption and design cost, this paper presents a custom-designed co-processor to implement OPC. The standard SRAM interface of the co-processor allows it to be controlled by various micro-controllers. The co-processor has been implemented in standard-cell based approach and manufactured in 0.18μm UMC technology. Tested with a 358-state 3-mixture 27-feature 800-word HMM, the co-processor operates at 10MHz to meet real-time requirement. The power consumption of this co-processor is 1.6mW, and the die size is 1.18mm2.

AB - In the CHMM (Continuous Hidden Markov Model) based speech recognition algorithm, Output Probability Calculation (OPC) is the most computation- intensive part. To reduce power consumption and design cost, this paper presents a custom-designed co-processor to implement OPC. The standard SRAM interface of the co-processor allows it to be controlled by various micro-controllers. The co-processor has been implemented in standard-cell based approach and manufactured in 0.18μm UMC technology. Tested with a 358-state 3-mixture 27-feature 800-word HMM, the co-processor operates at 10MHz to meet real-time requirement. The power consumption of this co-processor is 1.6mW, and the die size is 1.18mm2.

UR - http://www.scopus.com/inward/record.url?scp=70350180004&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=70350180004&partnerID=8YFLogxK

U2 - 10.1109/ISCAS.2009.5117762

DO - 10.1109/ISCAS.2009.5117762

M3 - Conference contribution

AN - SCOPUS:70350180004

SN - 9781424438280

T3 - Proceedings - IEEE International Symposium on Circuits and Systems

SP - 369

EP - 372

BT - 2009 IEEE International Symposium on Circuits and Systems, ISCAS 2009

T2 - 2009 IEEE International Symposium on Circuits and Systems, ISCAS 2009

Y2 - 24 May 2009 through 27 May 2009

ER -

Design a co-processor for output probability calculation in speech recognition

Abstract

Publication series

Other

Access

OpenUrl availability

Other files and links

Fingerprint

Cite this