DIGIT BOUNDARY AND RECOGNITION FOR MALAY ISOLATED DIGITS
Keywords:
HMM, DTW, zero crossing technique, log energy, word bounderAbstract
This paper proposes a speech recognition algorithm for Malay digits from 0 to 9. This system consistsof speech processing inclusive of digit boundary and recognition which uses zero crossing and energytechniques. Mel-Frequency Cepstral Coefficients (MFCC) vectors are used to provide an estimate ofthe vocal tract filter. Meanwhile dynamic time warping (DTW) is used to detect the nearest recordedvoice with appropriate global constraint is to set a valid search region because the variation of thespeech rate of the speaker is considered to be limited in a reasonable range, which means that it canprune the unreasonable search space. The algorithm is tested on speech samples that are recorded asa part of a Malay corpus. The results show that the algorithm managed to recognize almost 90.5% ofthe Malay digits for all recorded words.
References
Al-Haddad, S.A.R., Samad, S.A., and Hussain,A. (2006a). Automatic digit boundarysegmentation recognition. Proceedings ofMMU International Symposium onInformation and Communication Technology(M2USIC) 2006; Nov 16-17, 2006;Petaling Jaya, Selangor, Malaysia,p. 280-283.
Al-Haddad, S.A.R., Samad, S.A., and Hussain,A. (2006b). Automatic segmentation forMalay speech recognition. Proceedings ofPostgraduate Research Seminar 2006; Aug29-30, 2006, Bangi, Selangor, Malaysia,p. 235-239.
Analog Devices Inc. (1992). Digital SignalProcessing Applications Using the ADSP-2100 Family, Vol. 2. Prentice-Hall Inc.,Englewood Cliffs, NJ, USA, 608p.
Britannica, Encyclopedia Britannica Online.(2007). http://www.britannica.com/eb/article-9050292. Accessed date: Aug 8,2007.
Deng, L. and Huang, X. (2004). Challenges inadopting speech recognition. Communicationsof the ACM, 47(1):69-75.
European Telecommunications StandardsInstitute (ESTI). (2002). Speech Processing,Transmission and Quality Aspects(STQ); Distributed Speech Recognition;Advanced Front-end Feature ExtractionAlgorithm; Compression Algorithm. ETSIStandard Document - ES 201 108, SophiaAntipolis, France.
Gold, B. and Morgan, N. (2000). Speech andAudio Signal Processing. 1st ed. John Wileyand Sons, NY, USA, 537p.
Kim, D.S., Lee, S.Y., and Kil, R.M. (1999).Auditory processing of speech signals forrobust speech recognition in real worldnoisy environments. IEEE Trans. Speechand Audio Proc., 7(1):55-69.
Le, A. (2003). Rich transcription 2003: Springspeech-to-text transcription evaluationresults. Proc. RT03 Workshop, 2003; May19-20, 2003, Boston, MA, USA. Availablefrom: http://www.nist.gov/speech/tests/rt/rt2003/spring/presentations/rt03s-sttresults-v9.pdf. Accessed date: Oct 25, 2007.
Le, A., Fiscus, J., Garofolo, J., Przybocki, M.,Martin, A., Sanders, G., and Pallet, D. (2007).The 2002 NIST RT evaluation speech-totextresults. Proc. RT02 Workshop; May7-8, 2002; Vienna, Va, USA. Available from:http://www.nist.gov/speech/tests/rt/rt2002/presentations/rt02_stt_results_v5.pdf. Accessed date: Oct 25, 2007.
Milner, B.P. and Shao, X. (2002). Speech reconstructionfrom mel-frequency cepstralcoefficients using a source-filter model.Proceedings of the7th International Conferenceon Spoken Language Processing(ICSLP) 2002; Sept 16-20, 2002, DenverColorado, USA, p. 2,421-2,424.
Rabiner, L.R. and Sambur, M.R. (1975). Analgorithm for determining the endpoints ofisolated utterances. The Bell System TechnicalJ., 54(2):297-315.
Rabiner, L.R. and Schafer, R.W. (1978). DigitalProcessing of Speech Signals. 1st ed.Prentice-Hall Inc., Englewood Cliffs, NJ,USA, 509p.
Sheikh, H.S., Hong, K.S., and Tan, T.S. (2002).Design and development of speechcontrolrobotic manipulator arm. Proceedingsof the 7th International Conference onControl Automation, Robotics And Vision(ICARCV 2002; Dec 2-5, 2002; Singapore,p. 459-463.
