Chapter 10: Exercises
- (**)
pºâ¤¤¤å¥ÀµÀWÃФ§¤@:
½Ð¼g¤@Ó MATLAB µ{¦¡ vowelSpectrumPlot01.m¡A§¹¦¨¤U¦C¥\¯à¡G
- ¶i¦æ¤T¬í¿ýµ¡A¿ýµ³W®æ¬O 16KHz/16Bits/Mono¡A¿ýµ¤º®e¬O¤¤¤å¥Àµ¡u£¸¡v¡C¡]½ÐºÉ¶qºû«ùéwªºµ°ª¤Îµ¶q¡A¥i¥H¸ÕÅ¥¦¹½d¨ÒÀɮסC¡^
- ¨Ï¥Î buffer2.m ¤Î frame2volume.m ¨Ó¤Á¥Xµ®Ø¨Ãpºâµ¶q¡A¬ÛÃö³W®æ¬O frameSize=32ms, overlap=0ms¡C¡]´£¥Ü¡G§A¥²¶·¥ý±N frameSize ¤Î overlap ÂনÂI¼Æ¡C¡^½Ð§ì¥Xµ¶q³Ì¤jªºµ®Ø¡A¦b¨ä«e«á¦U§ì¥X 2 Óµ®Ø¡AÁ`¦@¦³ 5 Óµ®Ø¡C
- ¨Ï¥Î fft ¨Ópºâ³o¤Óµ®ØªºÀWÃСA§t¤j¤p©M¬Û¦ì¡A¨Ãµe¥X©Ò¦³¬ÛÃö¹Ï§Î¡G
- subplot(3,1,1)¡G¤Óµ®Øªº°T¸¹¹Ï¡C
- subplot(3,1,2)¡G¤Óµ®Øªº³æÃäÀWÃбj«×¹Ï¡C
- subplot(3,1,3)¡G¤Óµ®Øªº³æÃäÀWÃЬۦì¹Ï¡C
µe¥Xªº¹Ï§ÎÀ³¸ÓÃþ¦ü¤U¹Ï¡G
¡]´£¥Ü¡G(a)¹ïµ®Ø¶i¦æ fft ®É¡A¥i¥H¥ý¼¤W Hamming window¡A¥i¥HÅýºâ¥X¨ÓªºÀWÃФñ¸û°®²b¡Cµ{¦¡½X¬O¡Gframe=frame.*hamming(length(frame))¡C(b)¦bµe¥X³æÃäÀWÃбj«×¹Ï¡A¥i¥H¨Ï¥Î log ¨è«×¡A¤ñ¸û®e©ö¬Ý¥X¿Óªiµ²ºc¡A¤èªk¬O¦bµe²Ä¤Gӹϫá¡A¤U¹F¤U¦C©R¥O¡Gset(gca, 'yscale', 'log')¡C(c) ¬°¨Ï¬Û¦ì²£¥Í³sÄòªº¦±½u¡A¥i¥H¨Ï¥Î unwarp «ü¥O¡C¡^
- ½ÐÆ[¹î¬Ý¬Ý©Òµe¥Xªº¤TӹϡA¨º¤@ӹϪº¤±ø¦±½uªº¬Û¦ü«×³Ì°ª¡H¡]´£¥Ü¡G¥Ñ©ó§Ú̪ºµoµ³£¬O¡u£¸¡v¡A¦]¦¹¬Û¦ü«×³Ì°ªªº¹Ï©Ò¥Îªº¯S¼x¡A´N¬Oéw¤£Åܪº¯S¼x¡A¥i¥Î©ó»yµ¿ëÃÑ¡C¡^
- (**)
pºâ¤¤¤å¥ÀµÀWÃФ§¤G:
½Ð¼g¤@Ó MATLAB µ{¦¡ vowelSpectrumPlot02.m¡A¥\¯à¦p¦P¤W¤@ÃD¡A¦ý±N¿ýµ¤º®e§ï¦¨¤¤¤å¥Àµ¡u£¸£½¡v¡C¦¹ÃD¥i¥Î¨Ó±´°Q¦bµ°ªÅܤƪº±¡ªp¤U¡AÀWÃЪºÅܤƱ¡ªp¡C
- (***)
pºâ¤¤¤å¥ÀµÀWÃШöi¦æ¤ÀÃþ:
½Ð¼g¤@Ó MATLAB µ{¦¡ vowelSpectrumRecogChinese01.m¡A§¹¦¨¤U¦C¥\¯à¡G
- ¶i¦æ¤¬í¿ýµ¡A¿ýµ³W®æ¬O 16KHz/16Bits/Mono¡A¿ýµ¤º®e¬O¤¤¤å¥Àµ¡u£«¡B£¸¡B£¹¡B£®¡B£¬¡v¡C¡]µoµ½ÐºÉ¶q¥Ã¡A¨Ã¦b¥Àµ¤§¶¡µy³\°±¹y¡A¥H«K«áÄò¶i¦æ¦Û°Ê¤Áµ¡A¥i¥H¸ÕÅ¥¦¹½d¨ÒÀɮסC¡^
- ¨Ï¥Î endPointDetect.m (in SAP Toolbox) ¨Ó§ä¥X³o¤Óµªº¶}©l©Mµ²§ô¦ì¸m¡C½Ð½Õ¾ã¬ÛÃöºÝÂI°»´ú°Ñ¼Æ¡A¨Ï±o§Aªºµ{¦¡¯à°÷¦Û°Ê¦a¥¿½T¤Á¥X³o¤Óµ¡C§A¥i¥H³]©w plotOpt=1¡A¥H«K¨Ï¥Î endPointDetect.m ¨Óµe¥XºÝÂI°»´úµ²ªG¡A¥¿½T¹Ï§ÎÃþ¦ü¤U¹Ï¡G
- ¨Ï¥Î buffer2.m ¨Ó¹ï³o¤Óµ¤Á¥Xµ®Ø¡]frameSize=32ms, overlap=0ms¡A¦ý§A¥²¶·¥ý±N frameSize ¤Î overlap ÂনÂI¼Æ¡^¨ÃpºâÀWÃбj«×¡A½Ð±N³o¤ÓµªºÀWÃбj«×µe¥X¨Ó¡A¨C¤@Ó¥Àµ¹ïÀ³¨ì m ±ø¦±½u¡Am ¬O¦¹¥Àµªºµ®ØӼơA½Ð¤ñ¸û¦¹ m ±ø¦±½uªº¬Û¦ü«×¡Cµe¥X¹Ï§ÎÀ³¸ÓÃþ¦ü¤U¹Ï¡G
- ½Ð¥Î knncLoo.m (in Machine Learning Toolbox) ¨Óºâ¥X¨Ï¥Î KNNC ±N¸ê®Æ¤À¦¨¤Ãþ¥Àµªº leave-one-out ¿ëÃѲv¡C¡]¥i¥H³]©w k = 1¡C¡^
- ½Ð±N¯S¼x¼Æ¥Ø¥Ñ 1 ¼W¥[¨ì 128¡Aµe¥X KNNC ªº leave-one-out ¿ëÃѲv¹ï¯S¼x¼Æ¥Øªº¦±½u¡Cµe¥X¹Ï§ÎÀ³¸ÓÃþ¦ü¤U¹Ï¡G
- ½Ð¥Î PCA ©Î LDA ±N³o¨ÇÀWÃбj«×¸ê®Æ§ë¼v¨ì¤G«×ªÅ¶¡¡A¨Ã±N¹Ï§Îµe¥X¨Ó¡AÆ[¹î¬Ý¬Ý¬O§_¦P¤@Ãþ¥Àµ¡A¨ä¸ê®ÆÂI¦³»E¶°¦b¤@°_ªº¶É¦V¡C
- (**)
pºâ^¤å¥ÀµÀWÃШöi¦æ¤ÀÃþ:
½Ð¼g¤@Ó MATLAB µ{¦¡ vowelSpectrumRecogEnglish01.m.m¡A¥\¯à¦p¦P¤W¤@ÃD¡A¦ý±N¿ýµ¤º®e§ï¦¨^¤å¥Àµ¡ui¡Be¡B Ë¡Bo¡Bu...¡v¡C
Audio Signal Processing and Recognition (µ°T³B²z»P¿ëÃÑ)