• 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) RVMÀ» ÀÌ¿ëÇÑ À½¼ºÀνıâÀÇ ±¸Çö
¿µ¹®Á¦¸ñ(English Title) Implementation of Speech Recognizer using Relevance Vector Machine
ÀúÀÚ(Author) ±èâ±Ù   °í½Ã¿µ   À̱¤¼®   Çã°­ÀΠ  Chang-Keun Kim   Si-Young Koh   Kwang-Seok Lee   Kang  
¿ø¹®¼ö·Ïó(Citation) VOL 11 NO. 08 PP. 1596 ~ 1603 (2007. 08)
Çѱ۳»¿ë
(Korean Abstract)
º» ³í¹®¿¡¼­´Â À½¼ºÀÎ½Ä ½Ã½ºÅÛÀ» ±¸ÇöÇÔ¿¡ ÀÖ¾î Áß¿äÇÑ Æ¯Â¡ ÆĶó¹ÌÅÍ¿Í ÇнÀ£¬ÀÎ½Ä ¾Ë°í¸®ÁòÀÇ ¼±ÅÃÀ» À§ÇÑ Á¦¾ÈÀ» Çϱâ À§ÇÏ¿© °¢°¢ ¼¼ °¡ÁöÀÇ ¹æ¹ýÀ» Á¶ÇÕÇÏ¿© ÀÎ½Ä ½ÇÇèÀ» ¼öÇàÇÏ°í °ËÅäÇÏ¿´´Ù. µÎ Á¾·ùÀÇ ½ÇÇèÀ» ÅëÇÏ¿© Çϵå¿þ¾î ÀåÄ¡·Î ±¸ÇöÇÒ °æ¿ì º¸´Ù È¿°úÀûÀÎ À½¼ºÀÎ½Ä ½Ã½ºÅÛÀ» Á¦¾ÈÇÑ´Ù. ù ¹ø°·Î´Â Ư¡ ÆĶó¹ÌÅÍÀÇ ¼º´ÉÀ» Æò°¡Çϱâ À§ÇÏ¿© ±âÁ¸ÀÇ MFCC¿Í MFCC¸¦ PCA¿Í ICA¸¦ ÀÌ¿ëÇÏ¿© Ư¡ °ø°£À» º¯È­½ÃŲ »õ·Î¿î Ư¡ ÆĶó¹ÌÅ͸¦ Á¦¾ÈÇÏ¿© ÃÑ 3Á¾·ùÀÇ Æ¯Â¡ÆĶó¹ÌÅÍ¿¡ ´ëÇÑ ÀÎ½Ä ½ÇÇèÀ» ¼öÇàÇÏ¿´À¸¸ç, µÎ ¹ø°·Î´Â ÇнÀµ¥ÀÌÅÍ ¼ö¿¡ µû¸¥ HMM, SVM, RVMÀÇ ÀÎ½Ä ¼º´ÉÀ» ½ÇÇèÇÏ¿´´Ù. ÀÌ»óÀÇ ½ÇÇè¿¡ ÀÇÇÏ¿© ICA¿¡ ÀÇÇÑ Æ¯Â¡ ÆĶó¹ÌÅÍ°¡ Ư¡ °ø°£»ó¿¡¼­ÀÇ ³ôÀº ¼±Çü ºÐº°¼º¿¡ ÀÇÇØ MFCC¿Í ºñ±³ÇÏ¿© Æò±Õ 1.5%ÀÇ ¼º´ÉÇâ»óÀ» È®ÀÎÇÒ ¼ö ÀÖ¾úÀ¸¸ç ÇнÀµ¥ÀÌÅÍÀÇ °¨¼Ò¿¡ µû¸¥ ÀνĽÇÇè¿¡¼­´Â HMM°ú ºñ±³ÇÏ¿© RVM¿¡¼­ ÃÖ°í 3.25%ÀÇ ¼º´ÉÇâ»óÀ» È®ÀÎÇÏ¿´´Ù. ÀÌ¿¡ ±Ù°ÅÇÏ¿© TI»çÀÇ DSP(TMS320C32)¸¦ »ç¿ëÇÏ¿© À½¼ºÀνı⸦ ±¸ÇöÇÏ¿© ½Ç½Ã°£À¸·Î ½ÇÇèÇÏ¿© ½Ã¹Ä·¹À̼ǰú ºñ±³ÇÏ¿´´Ù. ÀÌ¿Í °°Àº °á°ú·Î¼­ º» ³í¹®¿¡¼­ Á¦¾ÈÇÏ´Â À½¼ºÀνĽýºÅÛÀ» À§ÇÑ È¿°úÀûÀÎ ¹æ¹ýÀº ICA¸¦ ÀÌ¿ëÇÑ Æ¯Â¡ ÆĶó¹ÌÅ͸¦ ÃßÃâÇÏ°í RVMÀ» ÀÌ¿ëÇÏ¿© ÀνÄÀ» ¼öÇàÇÏ´Â °ÍÀ̶ó ÆÇ´ÜÇÑ´Ù.
¿µ¹®³»¿ë
(English Abstract)
In this paper, we experimented by three kind of method for feature parameter, training method and recognition algorithm of most suitable for speech recognition system and considered. We decided speech recognition system of most suitable through two kind of experiment after we make speech recognizer. First, we did an experiment about three kind of feature parameter to evaluate recognition performance of it in speech recognizer using existent MFCC and MFCC new feature parameter that change characteristic space using PCA and ICA. Second, we experimented recognition performance of HMM, SVM and RVM by studying data number. By an experiment until now, feature parameter by ICA showed performance improvement of average 1.5% than MFCC by high linear discrimination from characteristic space. RVM showed performance improvement of maximum 3.25% than HMM in an experiment by decrease of studying data. As such result, effective method for speech recognition system to propose in this paper derives feature parameters using ICA and run recognition using RVM.
Å°¿öµå(Keyword) Speech Recognition   HMM   SVM   RVM   ICA   PCA   MFCC   DSP  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå