Error analysis of automatic speech recognition using principal direction divisive partitioning

David McKoskey; Daniel Boley

doi:10.1007/3-540-45164-1_28

Error analysis of automatic speech recognition using principal direction divisive partitioning

David McKoskey, Daniel Boley

Computer Science and Engineering

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

Abstract

This paper describes an experiment performed using the Principal Direction Divisive Partitioning algorithm (Boley, 1998) in order to extract linguistic word error regularities from several sets of medical dictation data. For each of six physicians, two hundred finished medical dictations aligned with their corresponding automatic speech recognition output were clustered and the results analyzed for linguistic regularities between and within clusters. Sparsity measures indicated a good fit between the algorithm and the input data. Linguistic analysis of the output clusters showed evidence of systematic word recognition error for short words, function words, words with destressed vowels, and phonological confusion errors due to telephony (recording) bandwidth interference. No qualitatively significant distinctions between clusters could be made by examining word errors alone, but the results confirmed several informally held hypotheses and suggested several avenues of further investigation, such as the examination of word error contexts.

Original language	English (US)
Title of host publication	Machine Learning
Subtitle of host publication	ECML 2000 - 11th European Conference on Machine Learning, Proceedings
Editors	Ramon Lopez de Mantaras, Enric Plaza
Publisher	Springer Verlag
Pages	263-270
Number of pages	8
ISBN (Print)	9783540451648
DOIs	https://doi.org/10.1007/3-540-45164-1_28
State	Published - 2000
Event	11th European Conference on Machine Learning, ECML 2000 - Barcelona, Catalonia, Spain Duration: May 31 2000 → Jun 2 2000

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	1810
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Other

Other	11th European Conference on Machine Learning, ECML 2000
Country	Spain
City	Barcelona, Catalonia
Period	5/31/00 → 6/2/00

Bibliographical note

Funding Information:
This work was partially supported by NSF grant IIS-9811229.

Access

10.1007/3-540-45164-1_28

Fingerprint Dive into the research topics of 'Error analysis of automatic speech recognition using principal direction divisive partitioning'. Together they form a unique fingerprint.

Cite this

McKoskey, D., & Boley, D. (2000). Error analysis of automatic speech recognition using principal direction divisive partitioning. In R. L. de Mantaras, & E. Plaza (Eds.), Machine Learning: ECML 2000 - 11th European Conference on Machine Learning, Proceedings (pp. 263-270). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1810). Springer Verlag. https://doi.org/10.1007/3-540-45164-1_28

Error analysis of automatic speech recognition using principal direction divisive partitioning. / McKoskey, David; Boley, Daniel.

Machine Learning: ECML 2000 - 11th European Conference on Machine Learning, Proceedings. ed. / Ramon Lopez de Mantaras; Enric Plaza. Springer Verlag, 2000. p. 263-270 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1810).

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

McKoskey, D & Boley, D 2000, Error analysis of automatic speech recognition using principal direction divisive partitioning. in RL de Mantaras & E Plaza (eds), Machine Learning: ECML 2000 - 11th European Conference on Machine Learning, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 1810, Springer Verlag, pp. 263-270, 11th European Conference on Machine Learning, ECML 2000, Barcelona, Catalonia, Spain, 5/31/00. https://doi.org/10.1007/3-540-45164-1_28

McKoskey D, Boley D. Error analysis of automatic speech recognition using principal direction divisive partitioning. In de Mantaras RL, Plaza E, editors, Machine Learning: ECML 2000 - 11th European Conference on Machine Learning, Proceedings. Springer Verlag. 2000. p. 263-270. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/3-540-45164-1_28

McKoskey, David ; Boley, Daniel. / Error analysis of automatic speech recognition using principal direction divisive partitioning. Machine Learning: ECML 2000 - 11th European Conference on Machine Learning, Proceedings. editor / Ramon Lopez de Mantaras ; Enric Plaza. Springer Verlag, 2000. pp. 263-270 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{7dbdf4896d8b4051b96263229d9a425f,

title = "Error analysis of automatic speech recognition using principal direction divisive partitioning",

abstract = "This paper describes an experiment performed using the Principal Direction Divisive Partitioning algorithm (Boley, 1998) in order to extract linguistic word error regularities from several sets of medical dictation data. For each of six physicians, two hundred finished medical dictations aligned with their corresponding automatic speech recognition output were clustered and the results analyzed for linguistic regularities between and within clusters. Sparsity measures indicated a good fit between the algorithm and the input data. Linguistic analysis of the output clusters showed evidence of systematic word recognition error for short words, function words, words with destressed vowels, and phonological confusion errors due to telephony (recording) bandwidth interference. No qualitatively significant distinctions between clusters could be made by examining word errors alone, but the results confirmed several informally held hypotheses and suggested several avenues of further investigation, such as the examination of word error contexts.",

author = "David McKoskey and Daniel Boley",

note = "Funding Information: This work was partially supported by NSF grant IIS-9811229.; 11th European Conference on Machine Learning, ECML 2000 ; Conference date: 31-05-2000 Through 02-06-2000",

year = "2000",

doi = "10.1007/3-540-45164-1_28",

language = "English (US)",

isbn = "9783540451648",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "263--270",

editor = "{de Mantaras}, {Ramon Lopez} and Enric Plaza",

booktitle = "Machine Learning",

}

TY - GEN

T1 - Error analysis of automatic speech recognition using principal direction divisive partitioning

AU - McKoskey, David

AU - Boley, Daniel

N1 - Funding Information: This work was partially supported by NSF grant IIS-9811229.

PY - 2000

Y1 - 2000

N2 - This paper describes an experiment performed using the Principal Direction Divisive Partitioning algorithm (Boley, 1998) in order to extract linguistic word error regularities from several sets of medical dictation data. For each of six physicians, two hundred finished medical dictations aligned with their corresponding automatic speech recognition output were clustered and the results analyzed for linguistic regularities between and within clusters. Sparsity measures indicated a good fit between the algorithm and the input data. Linguistic analysis of the output clusters showed evidence of systematic word recognition error for short words, function words, words with destressed vowels, and phonological confusion errors due to telephony (recording) bandwidth interference. No qualitatively significant distinctions between clusters could be made by examining word errors alone, but the results confirmed several informally held hypotheses and suggested several avenues of further investigation, such as the examination of word error contexts.

AB - This paper describes an experiment performed using the Principal Direction Divisive Partitioning algorithm (Boley, 1998) in order to extract linguistic word error regularities from several sets of medical dictation data. For each of six physicians, two hundred finished medical dictations aligned with their corresponding automatic speech recognition output were clustered and the results analyzed for linguistic regularities between and within clusters. Sparsity measures indicated a good fit between the algorithm and the input data. Linguistic analysis of the output clusters showed evidence of systematic word recognition error for short words, function words, words with destressed vowels, and phonological confusion errors due to telephony (recording) bandwidth interference. No qualitatively significant distinctions between clusters could be made by examining word errors alone, but the results confirmed several informally held hypotheses and suggested several avenues of further investigation, such as the examination of word error contexts.

UR - http://www.scopus.com/inward/record.url?scp=84974691623&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84974691623&partnerID=8YFLogxK

U2 - 10.1007/3-540-45164-1_28

DO - 10.1007/3-540-45164-1_28

M3 - Conference contribution

AN - SCOPUS:84974691623

SN - 9783540451648

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 263

EP - 270

BT - Machine Learning

A2 - de Mantaras, Ramon Lopez

A2 - Plaza, Enric

PB - Springer Verlag

T2 - 11th European Conference on Machine Learning, ECML 2000

Y2 - 31 May 2000 through 2 June 2000

ER -