6-1 Introduction to End-Point Detection (端é??µæ¸¬ä»‹ç´¹)

[english][all]

(½Ðª`·N¡G¤¤¤åª©¥»¨Ã¥¼ÀH­^¤åª©¥»¦P¨B§ó·s¡I)

Slides for this chapter

¡uºÝÂI°»´ú¡v¡]End-point Detection¡A²ºÙ EPD¡^ªº¥Ø¼Ð¬O­n¨M©w­µ°T¶}©l©Mµ²§ôªº¦ì¸m¡A©Ò¥H¤S¥i¥HºÙ¬° Speech Detection ©Î¬O VAD (Voice Activity Detection)¡CºÝÂI°»´ú¦b­µ°T³B²z»P¿ëÃѤ¤¡A§êºt¤@­Ó­«­nªº¨¤¦â¡C

±`¨£ªººÝÂI°»´ú¤èªk»P¬ÛÃöªº¯S¼x°Ñ¼Æ¡A¥i¥H¤À¦¨¨â¤jÃþ¡G

  1. ®É°ì¡]Time Domain¡^ªº¤èªk¡G­pºâ¶q¤ñ¸û¤p¡A¦]¦¹¤ñ¸û®e©ö²¾´Ó¨ì­pºâ¯à¤O¸û®tªº·L¹q¸£¥­¥x¡C
    1. ­µ¶q¡G¥u¨Ï¥Î­µ¶q¨Ó¶i¦æºÝÂI°»´ú¡A¬O³Ì²³æªº¤èªk¡A¦ý¬O·|¹ï®ð­µ³y¦¨»~§P¡C¤£¦Pªº­µ¶q­pºâ¤è¦¡¤]·|³y¦¨ºÝÂI°»´úµ²ªGªº¤£¦P¡A¦Ü©ó¬O­þ¤@ºØ­pºâ¤è¦¡¤ñ¸û¦n¡A¨ÃµL©w½×¡A»Ý­n¾a¤j¶qªº¸ê®Æ¨Ó´ú¸Õ±oª¾¡C
    2. ­µ¶q©M¹L¹s²v¡G¥H­µ¶q¬°¥D¡A¹L¹s²v¬°»²¡A¥i¥H¹ï®ð­µ¶i¦æ¸ûºë±KªºÀË´ú¡C
  2. ÀW°ì¡]Frequency Domain¡^ªº¤èªk¡G­pºâ¶q¤ñ¸û¤j¡A¦]¦¹¤ñ¸ûÃø²¾´Ó¨ì­pºâ¯à¤O¸û®tªº·L¹q¸£¥­¥x¡C
    1. ÀWÃЪºÅܲ§¼Æ¡G¦³Án­µªºÀWÃÐÅܤƸû³W«ß¡AÅܲ§¼Æ¸û§C¡A¥i§@¬°§PÂ_ºÝÂIªº°ò·Ç¡C
    2. ÀWÃЪºEntropy¡G§Ú­Ì¤]¥i¥H¨Ï¥Î¨Ï¥Î Entropy ¹F¨ìÃþ¦ü¤W­zªº¥\¯à¡C

Hint
²³æ¦a»¡¡A­Y¥u¬O¹ïÁn­µªi§Î°µ¤@¨Ç¸û²³æªº¹Bºâ¡A´N¬OÄÝ©ó®É°ìªº¤èªk¡C¥t¤@¤è­±¡A¤Z¬O­n¥Î¨ì³Å¥ß¸­Âà´«¡]Fourier Transform¡^¨Ó²£¥ÍÁn­µªºÀWÃСA´N¬OÄÝ©óÀWÃЪº¤èªk¡C³oºØ¤Àªk±`³Q¥Î¨Ó¹ï­µ°T³Bªº¤èªk¶i¦æ¤ÀÃþ¡A¦ý¦³®É­Ô¦³¤@¨Ç¼Ò½k¦a±a¡C¦³Ãö©óÀWÃÐ¥H¤Î³Å¥ß¸­Âà´«¡A·|¦b«áÄòªº³¹¸`»¡©ú¡C

¿ù»~ªººÝÂI°»´ú¡A¦b»y­µ¿ëÃѤW·|³y¦¨¨âºØ®ÄÀ³¡G

¥H¤U¦U¤p¸`±N°w¹ï³o¨âÃþªººÝÂI°»´ú¤èªk¨Ó¤¶²Ð¡C
Audio Signal Processing and Recognition (­µ°T³B²z»P¿ëÃÑ)