Research Interests
I am working on
HMM-based recognizer (improving and application). Three applications from
HMM-based recognizers have been proposed in the papers.
1. Speech recognition
2. AF extraction
3. AF-based language
identification
For HMM-based speech
recognition, I tried to optimize the way of using AF information as a
question set for tree-based state tying [13]. While AF-based
question generation can be considered as a successful semi-automatic question
generation, AF information might not be available in some language.
Therefore, a fully automatic question generation has also been proposed in [20]. I am
currently interested in combining MFCC and PLP. Some example has been
proposed in [19].
For Thai speech
recognition, due to the language characteristic, integrating syllable
structure [5]
and pronunciation variation rules [4] in HMM has
shown some benefit.
For HMM-based AF
extraction, backing-off context & gender dependent HMM serves as a good
AF extractor [15].
Pros and cons of the system has been compared and discussed with SVM [18]. I am currently
trying to integrate more state-of-the-art speech recognition technology in
the AF extraction system and use these extracted AF as additional information
for speech recognition. Some example of using this AF information in speech
recognition has been shown in [17].
The recognized AF sequences from AF detectors can be used
not only for further mapping to higher level of linguistic units (phones or
words) but also for language identification task [16].
The concept of this system is to use an AF detector generates a sequence of
AF units where two languages have different order of AF unit sequence.
Professional Experience
2001-2004
·
Researched
and developed Thai Speech Recognition.
·
Built
Thai speech corpus tools (Automatic speech transcription, Wave segmentation,
PD distributor, etc).
2000-2001
·
Researched cross-language adaptation of
acoustic models from Japanese to Thai.
1996-1997
·
Prepared
and ran practical courses
PhD in Computer Science (MUSTER Group)
·
Won 3-years scholarship from SFI (the Science Foundation Ireland)
·
GPA:
4.00
·
Won
3-years scholarship from Monbusho (Japanese Educational Government)
B. Eng. Electrical Engineering
·
GPA:
3.14
·
Won
4-year scholarship from Toshiba-Thai
foundation (TTFD)
2006
[20]
Kanokphara, S. and Carson-Berndsen, J., “Phonetic
Question Generation Using Misrecognition,” In Proc. the Ninth International Conference on TEXT, SPEECH and
DIALOGUE (TSD), 407-414, September 2006.
[19]
Kanokphara, S. and Carson-Berndsen, J., “A Study Of Phone
Recognizer Combination For Higher Accuracy In TIMIT Phone Recognition,”
In Proc. the 9th Western Pacific Acoustics
Conference (WESPAC IX), 154-158, June 2006.
[18]
Kanokphara, S., Macek, J. and Carson-Berndsen,
J., “Comparative Study: HMM&SVM for Automatic Articulatory
Feature Extraction,” In Proc. the Nineteenth
International Conference on Industrial and Enginering
Applications of Artificial Intelligence and Expert Systems (IEA/AIE),
674-681, June 2006.
[17] Aioanei D., Carson-Berndsen, J.
and Kanokphara, S., “Diagnostic
Evaluation of Phonetic Feature Extraction Engines: A Case Study with the Time
Map Model,” In Proc. the Nineteenth
International Conference on Industrial and Enginering
Applications of Artificial Intelligence and Expert Systems (IEA/AIE),
691-700, June 2006.
[16]
Kanokphara, S. and Carson-Berndsen, J., “Articulatory-Acoustic-Feature-based
Automatic Language Identification,” In Proc. Tutorial and Research Workshop
(ITRW) on Multilingual Speech and Language Processing, paper 013, April
2006.
2005
[15]
Kanokphara, S. and Carson-Berndsen, J., “Backing-Off Context-
& Gender-Dependent Models for Better Articulatory Feature Extraction”,
In Proc. The Second International Conference
on Intelligent Computing and Information Systems (ICICIS), 287-292, 2005.
[14]
Kanokphara, S. and Carson-Berndsen, J., “Better HMM-Based
Articulatory Feature Extraction with Context-Dependent Model”,
In Proc. The 18th International
Florida Artificial Intelligence Research Society Conference (FLAIRS),
370-374, 2005.
[13]
Kanokphara, S. and Carson-Berndsen, J., “Feature-Table-Based
Automatic Question Generation for Tree-Based State Tying: A Practical
Implementation”, In Proc. The 18th International
Conference on Industrial & Engineering Applications of Artificial
Intelligence & Expert Systems (IEA/AIE05), 95-97, 2005.
[12] Kanokphara, S., Geumann,
A. and Carson-Berndsen, J., “Accessing Language
Specific Linguistic Information for Triphone Model Generation: Feature Tables
in a Speech Recognition System”, In Proc. The 2nd Language & Technology Conference:
Human Language Technologies as a Challenge for Computer Science and
Linguistics, 7-10.
[11] Macek, J., Kanokphara S.
and Geumann, A., “Articulatory-acoustic
Feature Recognition: Comparison of Machine Learning and HMM methods”,
In Proc. 10th International Conference
on Speech and Computer (SPECOM), 99-102, 2005.
2004
[10]
Kanokphara, S. Carson-Berndsen, J., "Automatic Question
Generation for HMM State Tying using a Feature Table", In Proc. Tenth Australian international
Conference on Speech Science & Technology (SST), 322-327, 2004.
[9] Kasuriya, S., Kanokphara, S., Thatphithakkul,
N., Cotsomrong, P.and Sunpethniyom, T., “Context-independent
Acoustic Models for Thai Speech Recognition”, In Proc. International Symposium on
Communications and Information Technologies 2004 (ISCIT), 904-908, 2004.
[8] Thatphithakkul, N., Kanokphara, S., “Hmm Parameter Optimization
using Tabu Search”, In Proc. International Symposium on
Communications and Information Technologies 2004 (ISCIT), 904-908, 2004.
2003
[7] Kanokphara, S., Tesprasit, V., Thongprasirt, R.
“Pronunciation
Variation Speech Recognition Without New Dictionary Construction”,
NECTEC Technical Journal,
13(4):317-321, 2003.
[6]
Kasuriya, S., Sornlertlamvanich,
V., Cotsomrong, P., Kanokphara, S., Thatphithakkul, N., “Thai Speech Corpus
for Speech Recognition”, In Proc. O-COCOSDA,
54-61, 2003.
[5] Kanokphara, S.,
“Syllable
Structure Based Phonetic Units for Context-Dependent Continuous Thai Speech Recognition”,In Proc. Eurospeech,
797-800, 2003.
[4] Kanokphara, S., Tesprasit, V., Thongprasirt, R.
“Pronunciation
Variation Speech Recognition Without Dictionary Modification on Sparse
Database”, In Proc. ICASSP, vol. 1, 764-767,
2003.
2002
[3] Wutiwiwatchai, C., Cotsomrong,
P., Suebvisai, S., Kanokphara, S., “Phonetically
Distributed Continuous Speech Corpus for Thai Language”,In Proc. LREC, vol. 3, 869-872, 2002.
[2] Thongprasirt, R., Sornlertlamvanich,
V., Cotsomrong, P., Suebvisai,
S., Kanokphara, S., “Progress Report on
Corpus Development and Speech Technology in Thailand”, In
Proc. SNLP-O-COCOSDA,
300-306, 2002.
[1]
Tarsaku, P., Kanokphara, S., “A Study of
HMM-based automatic segmentations for Thai Continuous Speech Recognition System”,In Proc. SNLP-O-COCOSDA,
217-220, 2002.
Computing:
·
Familiar
with both UNIX and Windows based OS
·
Advanced
skill in C/C++, OOP
and design pattern
·
HTK
advanced user
Leadership/Teamwork:
·
Vice-president
of Movie Club, NECTEC
·
Head
of Staff Department, Cheer Club, Thammasat University
·
Member
of Tai Chi club in UCD
Languages: Thai, English, Japanese
|
