[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
- ®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
- 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
- µ¶q©M¹L¹s²v¡G¥Hµ¶q¬°¥D¡A¹L¹s²v¬°»²¡A¥i¥H¹ï®ðµ¶i¦æ¸ûºë±KªºÀË´ú¡C
- ÀW°ì¡]Frequency Domain¡^ªº¤èªk¡Gpºâ¶q¤ñ¸û¤j¡A¦]¦¹¤ñ¸ûÃø²¾´Ó¨ìpºâ¯à¤O¸û®tªº·L¹q¸£¥¥x¡C
- ÀWÃЪºÅܲ§¼Æ¡G¦³ÁnµªºÀWÃÐÅܤƸû³W«ß¡AÅܲ§¼Æ¸û§C¡A¥i§@¬°§PÂ_ºÝÂIªº°ò·Ç¡C
- ÀWÃЪºEntropy¡G§Ṳ́]¥i¥H¨Ï¥Î¨Ï¥Î Entropy ¹F¨ìÃþ¦ü¤Wzªº¥\¯à¡C
¿ù»~ªººÝÂI°»´ú¡A¦b»yµ¿ëÃѤW·|³y¦¨¨âºØ®ÄÀ³¡G
¥H¤U¦U¤p¸`±N°w¹ï³o¨âÃþªººÝÂI°»´ú¤èªk¨Ó¤¶²Ð¡C
- False Rejection¡G±N Speech »~»{¬° Silence/Noise¡A¦]¦Ó³y¦¨µ°T¿ëÃѲv¤U°
- False Acceptance¡G±N Silence/Noise »~»{¬° Speech¡A¦¹®Éµ°T¿ëÃѲv¤]·|¤U°¡A¦ý¬O§ÚÌ¥i¥H¦b³]p¿ëÃѾ¹®É¡A«e«á¥[¤W¥i¯àªºÀRµÁn¾Ç¼Ò«¬¡A¦¹®É¿ëÃѲvªº¤U°´N·|¤ñ«eªÌ¨Óªº©M½w¡C
Audio Signal Processing and Recognition (µ°T³B²z»P¿ëÃÑ)