• Àüü
  • ÀüÀÚ/Àü±â
  • Åë½Å
  • ÄÄÇ»ÅÍ
´Ý±â

»çÀÌÆ®¸Ê

Loading..

Please wait....

±¹³» ³í¹®Áö

Ȩ Ȩ > ¿¬±¸¹®Çå > ±¹³» ³í¹®Áö > Çѱ¹Á¤º¸Åë½ÅÇÐȸ ³í¹®Áö (Journal of the Korea Institute of Information and Communication Engineering)

Çѱ¹Á¤º¸Åë½ÅÇÐȸ ³í¹®Áö (Journal of the Korea Institute of Information and Communication Engineering)

Current Result Document :

ÇѱÛÁ¦¸ñ(Korean Title) Matrix FactorizationÀ» ÀÌ¿ëÇÑ À½¼º Ư¡ ÆĶó¹ÌÅÍ ÃßÃâ ¹× ÀνÄ
¿µ¹®Á¦¸ñ(English Title) Feature Parameter Extraction and Speech Recognition Using Matrix Factorization
ÀúÀÚ(Author) À̱¤¼®   Çã°­ÀΠ  Kwang-Seok Lee   Kang-In Hur  
¿ø¹®¼ö·Ïó(Citation) VOL 10 NO. 07 PP. 1307 ~ 1311 (2006. 07)
Çѱ۳»¿ë
(Korean Abstract)
º» ¿¬±¸¿¡¼­´Â Çà·Ä ºÐÇØ(Matrix Factorization)¸¦ ÀÌ¿ëÇÏ¿© À½¼º ½ºÆåÆ®·³ÀÇ ºÎºÐÀû Ư¡À» ³ªÅ¸³¾ ¼ö ÀÖ´Â »õ·Î¿î À½¼º ÆĶó¹ÌÅ͸¦ Á¦¾ÈÇÑ´Ù. Á¦¾ÈµÈ ÆĶó¹ÌÅÍ´Â Çà·Ä³»ÀÇ ¸ðµç ¿ø¼Ò°¡ À½¼ö°¡ ¾Æ´Ï¶ó´Â Á¶°Ç¿¡¼­ Çà·ÄºÐÇØ °úÁ¤À» °ÅÄ¡°Ô µÇ°í °íÂ÷¿øÀÇ µ¥ÀÌÅÍ°¡ È¿°úÀûÀ¸·Î Ãà¼ÒµÇ¾î ³ªÅ¸³²À» ¾Ë¼ö ÀÖ´Ù. Â÷¿ø Ãà¼ÒµÈ µ¥ÀÌÅÍ´Â ÀÔ·Â µ¥ÀÌÅÍÀÇ ºÎºÐÀûÀΠƯ¼ºÀ» Ç¥ÇöÇÑ´Ù. À½¼º Ư¡ ÃßÃâ °úÁ¤¿¡¼­ ÀϹÝÀûÀ¸·Î »ç¿ëµÇ´Â ¸á ÇÊÅ͹ðÅ©(Mel-FilterBank)ÀÇ Ãâ·ÂÀ» Non-Negative Çà ·Ä ºÐÇØ (NMF: Non-Negative Matrix Factorization) ¾Ë°í¸® ÁòÀÇ ÀÔ ·Â À¸·Î »ç¿ëÇÏ°í£¬¾Ë°í¸® ÁòÀ» ÅëÇØ Â÷¿ø Ãà¼ÒµÈ µ¥ÀÌÅ͸¦ À½¼ºÀνıâÀÇ ÀÔ·ÂÀ¸·Î »ç¿ëÇÏ¿© ¸á ÁÖÆļö ĸ½ºÆ®·³ °è¼ö (MFCC: Mel Frequency Cepstral Coefficient)ÀÇ Àνİá°ú¿Í ºñ±³ÇØ º¸¾Ò´Ù. Àνİá°ú¸¦ ÅëÇÏ¿© ÀϹÝÀûÀ¸·Î À½¼ºÀνıâÀÇ ¼º´ÉÆò°¡¸¦ À§ÇØ »ç¿ëµÇ´Â MFCC¿¡ ºñÇÏ¿© Á¦¾ÈµÈ Ư¡ ÆĶó¹Ì ÅÍ °¡ ÀÎ½Ä ¼º´ÉÀÌ ¶Ù ¾î ³²À» ¾Ë ¼ö ÀÖ¾ú´Ù.
¿µ¹®³»¿ë
(English Abstract)
In this paper, we propose new speech feature parameter using the Matrix Factorization for appearance part-based features of speech spectrum. The proposed parameter represents effective dimensional reduced data from multi-dimensional feature data through matrix factorization procedure under all of the matrix elements are the non-negative constraint. Reduced feature data presents part-based features of input data. We verify about usefulness of NMF(Non-Negative Matrix Factorization) algorithm for speech feature extraction applying feature parameter that is got using NMF in Mel-scaled filter bank output. According to recognition experiment results, we confirm that proposed feature parameter is superior to MFCC(Mel-Frequency Cepstral Coefficient) in recognition performance that is used generally.
Å°¿öµå(Keyword) Speech Feature Extraction   Non-Negative Matrix Factorization   Mel-scaled Filter Bank  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå