[Video at Xiph] Digital Show and Tell - Czech subtitles

[mkmk]

Hello Xiph.Org,

I liked your "Digital Show and Tell" video which I found highly educative 
and also fun to watch,
since I'm interested in these topics, so I decided to translate the 
subtitles to Czech language for you.
Attached please find subtitles in VTT format (file is based on your EN 
subtitles).
I hope my subtitles will appear online and help someone from Czech Republic 
with understanding your video.
BR
Michal Kosek
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.xiph.org/pipermail/video/attachments/20130306/c13ecd84/attachment-0001.htm 
-------------- next part --------------
WEBVTT

1
00:00:08.252 --> 00:00:11.550
Ahoj, j? jsem Monty Montgomery
ze spole?nost? Red Hat a Xiph.Org.

2
00:00:11.550 --> 00:00:18.430
Ned?vno jsem napsal ?l?nek o digit?ln?m zvuku a tom,
pro? nem? smysl stahovat hudbu v kvalit? 24bit?/192kHz.

3
00:00:18.430 --> 00:00:23.433
V ?l?nku jsem se zm?nil o tom,
?e digit?ln? sign?l nem? schodov? pr?b?h

4
00:00:23.433 --> 00:00:28.680
a konverz? na analogov?
rozhodn? schody nez?sk?te.

5
00:00:29.865 --> 00:00:33.865
Pr?v? <b>tohle</b> bylo hlavn? t?ma,
o kter?m lid? za?ali diskutovat.

6
00:00:33.865 --> 00:00:37.221
V?c ne? polovina email?
obsahovala dotazy a p?ipom?nky

7
00:00:37.221 --> 00:00:39.663
k chov?n? digit?ln?ho sign?lu.

8
00:00:39.894 --> 00:00:45.285
Kdy? v?s to tolik zaj?m?, pohraji si s n?jak?mi
jednoduch?mi digit?ln?mi sign?ly.

9
00:00:49.747 --> 00:00:51.006
P?edpokl?dejme, ?e netu??me,

10
00:00:51.006 --> 00:00:54.089
jak se digit?ln? sign?l chov?.

11
00:00:54.734 --> 00:00:56.841
V tom p??pad? pro n?s nem? smysl

12
00:00:56.841 --> 00:00:59.049
testovat na digit?ln?ch p??stroj?ch.

13
00:00:59.049 --> 00:01:00.937
Na?t?st? st?le existuje

14
00:01:00.937 --> 00:01:04.020
mnoho funk?n?ch analogov?ch p??stroj?.

15
00:01:04.020 --> 00:01:05.972
Nejd??ve pot?ebujeme gener?tor sign?lu,

16
00:01:05.972 --> 00:01:08.190
z kter?ho z?sk?me analogov? sign?l.

17
00:01:08.190 --> 00:01:12.692
V na?em p??pad? to je HP3325 z roku 1978.

18
00:01:12.692 --> 00:01:14.153
Je to st?le dobr? gener?tor.

19
00:01:14.153 --> 00:01:15.614
Pokud v?m nevad? jeho rozm?ry,

20
00:01:15.614 --> 00:01:16.532
hmotnost,

21
00:01:16.532 --> 00:01:17.577
spot?eba

22
00:01:17.577 --> 00:01:18.910
a hlu?n? v?tr?k,

23
00:01:18.910 --> 00:01:20.329
d? se koupit na eBayi.

24
00:01:20.329 --> 00:01:23.863
N?kdy i za nepatrn? v?c, ne? je po?tovn?.

25
00:01:24.617 --> 00:01:28.500
Analogov? pr?b?hy budeme sledovat
na analogov?m osciloskopu.

26
00:01:28.500 --> 00:01:31.550
Toto je Tektronix 2246 z devades?t?ch let.

27
00:01:31.550 --> 00:01:34.761
Jeden z posledn?ch a tak? nejlep??ch
analogov?ch osciloskop?.

28
00:01:34.761 --> 00:01:36.807
Nem?l by chyb?t v ??dn? dom?c? d?ln?.

29
00:01:37.716 --> 00:01:40.852
Frekven?n? spektrum budeme pozorovat

30
00:01:40.852 --> 00:01:43.177
na analogov?m spektr?ln?m analyz?toru.

31
00:01:43.177 --> 00:01:47.732
Tento HP3585 je ze stejn? produktov?
?ady jako n?? gener?tor.

32
00:01:47.732 --> 00:01:50.615
Stejn? jako ostatn? p??stroje obsahuje omezen?

33
00:01:50.615 --> 00:01:52.905
a komicky rozm?rn? mikrokontrol?r,

34
00:01:52.905 --> 00:01:56.276
ale cesta sign?lu od vstupu a? na obrazovku

35
00:01:56.276 --> 00:01:58.537
je v?hradn? analogov?.

36
00:01:58.537 --> 00:02:00.329
V?echny tyto p??stroje jsou zastaral?,

37
00:02:00.329 --> 00:02:01.993
ale (krom jejich ton??e)

38
00:02:01.993 --> 00:02:03.844
maj? st?le v?born? parametry.

39
00:02:04.536 --> 00:02:06.868
Pr?v? generujeme sinusovku

40
00:02:06.868 --> 00:02:12.829
s kmito?tem 1kHz, 1V efektivn?.

41
00:02:13.414 --> 00:02:15.220
Na osciloskopu vid?me sinusovku,

42
00:02:15.220 --> 00:02:21.428
kter? m? skute?n? 1kHz a 1V efektivn?,

43
00:02:21.428 --> 00:02:24.108
co? odpov?d? amplitud? 2.8V.

44
00:02:24.308 --> 00:02:27.561
A to odpov?d? i m??en? na spektr?ln?m analyz?toru,

45
00:02:27.561 --> 00:02:30.644
kter? tak? ukazuje b?l? ?um n?zk? hladiny

46
00:02:30.644 --> 00:02:32.190
a men?? harmonick? zkreslen?,

47
00:02:32.190 --> 00:02:36.649
s nejvy???m vrcholem asi 70dB pod ?rovn? sign?lu.

48
00:02:36.649 --> 00:02:38.612
V na?em p??pad? je bezv?znamn?,

49
00:02:38.612 --> 00:02:40.574
ale cht?l jsem na n?j upozornit,

50
00:02:40.574 --> 00:02:42.452
pro p??pad ?e byste si jej nev?imli. 

51
00:02:44.036 --> 00:02:47.142
A te? p?id?me digit?ln? vzorkov?n?. 

52
00:02:48.557 --> 00:02:51.024
Pou?ijeme b??nou

53
00:02:51.024 --> 00:02:53.374
zvukovou kartu eMagic USB1.

54
00:02:53.374 --> 00:02:55.337
Je p?es deset let star?

55
00:02:55.337 --> 00:02:57.257
a d?vno p?ekonan?.

56
00:02:57.964 --> 00:03:02.676
Sou?asn? konvertor lehko p?ekon? jej? parametry.

57
00:03:03.076 --> 00:03:07.924
Plochost, linearita,
jitter, ?um, prost? v?echno...

58
00:03:07.924 --> 00:03:09.353
...co z?ejm? nepost?ehnete.

59
00:03:09.353 --> 00:03:11.604
?e n?co m??eme zm??it neznamen?,

60
00:03:11.604 --> 00:03:13.609
?e jsme schopni to sly?et.

61
00:03:13.609 --> 00:03:16.404
A u? tyhle star? krabi?ky

62
00:03:16.404 --> 00:03:18.643
byly t?m?? ide?ln? transparentn?.

63
00:03:20.244 --> 00:03:22.825
eMagic zapoj?m do sv?ho ThinkPadu,

64
00:03:22.825 --> 00:03:26.121
kter? zobrazuje digit?ln?
pr?b?h sign?lu a jeho spektrum.

65
00:03:26.121 --> 00:03:28.788
Pak po?le sign?l zp?t do eMagicu,

66
00:03:28.788 --> 00:03:30.921
kde je znovu p?eveden do analogov? podoby

67
00:03:30.921 --> 00:03:33.332
a v?stup je zobrazen na p??stroj?ch.

68
00:03:33.332 --> 00:03:35.582
Vstup nalevo, v?stup napravo.

69
00:03:40.211 --> 00:03:41.214
Jdeme na to!

70
00:03:41.214 --> 00:03:43.924
Za?neme konverz? analogov?ho sign?lu na digit?ln?

71
00:03:43.924 --> 00:03:47.347
a zp?t na analogov? s ??dn?mi mezikroky.

72
00:03:47.347 --> 00:03:49.268
Gener?tor sign?lu generuje sinusovku

73
00:03:49.268 --> 00:03:52.649
s frekvenc? 1kHz, stejn? jako p?edt?m.

74
00:03:52.649 --> 00:03:57.428
Osciloskop na vstupu ukazuje analogovou sinusovku.

75
00:03:57.428 --> 00:04:01.694
Sign?l p?evedeme pulzn? k?dovou modulac? na 16bit?/44.1kHz,

76
00:04:01.694 --> 00:04:03.828
stejn? jako na kompaktn?m disku.

77
00:04:03.828 --> 00:04:07.156
Spektrum digitalizovan?ho sign?lu
odpov?d? tomu, co jsme vid?li d??ve...

78
00:04:07.156 --> 00:04:10.836
...a co vid?me na spektr?ln?m analyz?toru,

79
00:04:10.836 --> 00:04:15.154
pomineme-li vy??? ?um
vysokoimpedan?n?ho vstupu.

80
00:04:15.154 --> 00:04:15.956
Nyn?

81
00:04:18.248 --> 00:04:20.798
zobrazen? pr?b?hu ukazuje digitalizovanou sinusovku

82
00:04:20.798 --> 00:04:23.966
jako schodov? graf, jeden schod na ka?d? vzorek.

83
00:04:23.966 --> 00:04:26.388
Pokud se pod?v?me na v?stupn? sign?l,

84
00:04:26.388 --> 00:04:29.054
kter? byl p?eveden z digit?lu zp?t na analog,

85
00:04:29.054 --> 00:04:32.052
vid?me, ?e je toto?n? s p?vodn? sinusovkou.

86
00:04:32.052 --> 00:04:33.483
??dn? schody.

87
00:04:33.914 --> 00:04:37.193
1kHz je st?le dost n?zk? kmito?et,

88
00:04:37.193 --> 00:04:40.633
Mo?n? jsou schody jen t??ko
post?ehnuteln?, nebo byly vyhlazeny.

89
00:04:40.739 --> 00:04:49.492
Zvol?me tedy vy??? frekvenci
bli??? Nyquistov?, ?ekn?me 15kHz.

90
00:04:49.492 --> 00:04:53.545
Sinusovka je te? zastoupen?
m?n? ne? t?emi vzorky na periodu

91
00:04:53.545 --> 00:04:55.838
a digit?ln? pr?b?h nevypad? dob?e.

92
00:04:55.838 --> 00:04:59.798
Vzhled klame. Na analogov?m v?stupu...

93
00:05:01.876 --> 00:05:06.033
je bezchybn? sinusovka, stejn? jako p?vodn?.

94
00:05:06.633 --> 00:05:09.228
Pokra?ujme d?l.

95
00:05:17.353 --> 00:05:20.151
16kHz....

96
00:05:23.198 --> 00:05:25.616
17kHz...

97
00:05:28.201 --> 00:05:29.945
18kHz...

98
00:05:33.822 --> 00:05:35.548
19kHz...

99
00:05:40.457 --> 00:05:42.465
20kHz.

100
00:05:49.097 --> 00:05:52.350
V?tejte na horn? hranici schopnost? lidsk?ho ucha.

101
00:05:52.350 --> 00:05:54.377
V?stupn? sign?l je st?le dokonal?.

102
00:05:54.377 --> 00:05:58.025
??dn? rozt?epen? okraje,
??dn? hrany, ??dn? schody.

103
00:05:58.025 --> 00:06:01.342
Kam se tedy schody pod?ly?

104
00:06:01.342 --> 00:06:03.198
Nesna?te se odpov?d?t, je to chyt?k.

105
00:06:03.198 --> 00:06:04.318
Nikdy tam nebyly.

106
00:06:04.318 --> 00:06:06.652
Vykreslovat digit?ln? pr?b?h jako schody...

107
00:06:08.712 --> 00:06:10.772
je samo o sob? chybou.

108
00:06:10.942 --> 00:06:11.998
Pro??

109
00:06:11.998 --> 00:06:14.366
Schody jsou funkc? spojit?ho ?asu.

110
00:06:14.366 --> 00:06:16.201
Jsou zubat? a definovan? po ??stech,

111
00:06:16.201 --> 00:06:19.700
ale v ka?d?m bod? v ?ase maj? hodnotu.

112
00:06:19.700 --> 00:06:22.004
Vzorkovan? sign?l je ?pln? jin?.

113
00:06:22.004 --> 00:06:23.337
Je v ?ase diskr?tn?.

114
00:06:23.337 --> 00:06:27.337
M? hodnotu jen v okam?iku m??en? ka?d?ho vzorku

115
00:06:27.337 --> 00:06:32.596
a v?ude jinde nen? definov?n -- nem? hodnotu.

116
00:06:32.596 --> 00:06:36.666
V ?ase diskr?tn? sign?l
by se m?l kreslit jako bodov? graf. 

117
00:06:40.020 --> 00:06:42.974
Odpov?daj?c? spojit? analogov? sign?l

118
00:06:42.974 --> 00:06:45.364
prob?h? hladce ka?d?m bodem

119
00:06:45.364 --> 00:06:50.153
a to plat? pro v?echny frekvence.

120
00:06:50.153 --> 00:06:53.033
Zaj?mav?m a ne zcela z?ejm?m faktem je,

121
00:06:53.033 --> 00:06:55.454
?e existuje pr?v? jeden frekven?n? omezen? sign?l,

122
00:06:55.454 --> 00:06:57.417
kter? prob?h? v?emi vzorky.

123
00:06:57.417 --> 00:06:58.708
?e?en? je jednozna?n?.

124
00:06:58.708 --> 00:07:01.246
Pokud vzorkujete frekven?n? omezen? sign?l

125
00:07:01.246 --> 00:07:02.612
a p?evedete ho zp?t na analogov?,

126
00:07:02.612 --> 00:07:06.462
jedin? mo?n? v?stup je p?vodn? sign?l.

127
00:07:06.462 --> 00:07:07.838
Je?t? ne? ?eknete:

128
00:07:07.838 --> 00:07:11.721
?J? ale mohu nakreslit i jin? sign?l, kter? prob?h? v?emi body.?

129
00:07:11.721 --> 00:07:14.283
Ano, to opravdu m??ete, ale...

130
00:07:17.268 --> 00:07:20.521
pokud se jakkoliv odchyluje od origin?lu,

131
00:07:20.521 --> 00:07:24.905
pak obsahuje frekvenci vy??? nebo rovnou Nyquistov?,

132
00:07:24.905 --> 00:07:26.185
??m? poru?uje podm?nku frekven?n?ho omezen?

133
00:07:26.185 --> 00:07:28.358
a nen? platn?m ?e?en?m.

134
00:07:28.574 --> 00:07:30.036
Tak pro? si v?ichni mysl?,

135
00:07:30.036 --> 00:07:32.702
?e digit?ln? sign?l jsou schody?

136
00:07:32.702 --> 00:07:34.900
Napadaj? m? dva d?vody:

137
00:07:34.900 --> 00:07:37.956
Za prv?: je jednoduch? p?ev?st

138
00:07:37.972 --> 00:07:39.294
vzorkovan? sign?l na schody

139
00:07:39.294 --> 00:07:42.409
prota?en?m hodnoty ka?d?ho
vzorku a? k tomu dal??mu.

140
00:07:42.409 --> 00:07:44.414
Tomu se ??k? extrapolace nult?ho ??du

141
00:07:44.414 --> 00:07:47.913
a je d?le?it?m krokem v n?kter?ch D/A p?evodn?c?ch,

142
00:07:47.913 --> 00:07:50.089
zejm?na v t?ch jednodu???ch.

143
00:07:50.089 --> 00:07:55.591
Tak?e ka?d?, kdo se zaj?m? a D/A p?evodn?ky,

144
00:07:55.592 --> 00:07:59.550
nejsp?? naraz? na schodov? graf.

145
00:07:59.550 --> 00:08:01.982
To ale nen? hotov? konverze

146
00:08:01.982 --> 00:08:04.250
a neodpov?d? v?stupn?mu sign?lu.

147
00:08:04.944 --> 00:08:05.684
Za druh?

148
00:08:05.684 --> 00:08:07.529
(a pravd?podobn?ji),

149
00:08:07.529 --> 00:08:09.449
in?en??i, kte?? v?ci rozum? l?pe,

150
00:08:09.449 --> 00:08:10.441
jako j?,

151
00:08:10.441 --> 00:08:13.193
kresl? schody, p?esto?e v?d?, ?e to nen? technicky spr?vn?.

152
00:08:13.193 --> 00:08:15.571
Je to jako jednorozm?rn? verze

153
00:08:15.571 --> 00:08:17.395
velk?ch pixel? v grafick?m editoru.

154
00:08:17.395 --> 00:08:19.241
Pixely tak? nejsou ?tvercov?,

155
00:08:19.241 --> 00:08:23.081
jsou to vzorky dvourozm?rn? funkce,

156
00:08:23.081 --> 00:08:26.366
tak?e jsou to tak? nekone?n? mal? body.

157
00:08:26.366 --> 00:08:28.500
Pracovat s ??mkoliv nekone?n? mal?m

158
00:08:28.500 --> 00:08:30.804
je ale osinou v zadku.

159
00:08:30.804 --> 00:08:32.212
Proto ty velk? ?tverce.

160
00:08:32.212 --> 00:08:35.966
Digit?ln? schody jsou to sam?.

161
00:08:35.966 --> 00:08:37.684
Je to jen praktick? kresba.

162
00:08:37.684 --> 00:08:40.404
Schody tam ve skute?nosti nejsou.

163
00:08:45.652 --> 00:08:48.233
Konverz? digit?ln?ho sign?lu na analogov?

164
00:08:48.233 --> 00:08:50.900
z?sk?me vyhlazen? pr?b?h,
nehled? na bitovou hloubku:

165
00:08:50.900 --> 00:08:53.193
24 nebo 16 bit?...

166
00:08:53.193 --> 00:08:54.196
nebo 8...

167
00:08:54.196 --> 00:08:55.486
na tom nez?le??.

168
00:08:55.486 --> 00:08:57.534
Znamen? to tedy, ?e po?et bit?

169
00:08:57.534 --> 00:08:58.953
je irelevantn??

170
00:08:59.245 --> 00:09:00.521
Rozhodn? ne.

171
00:09:02.121 --> 00:09:06.046
2. kan?l obsahuje stejnou sinusovku,

172
00:09:06.046 --> 00:09:09.086
kterou kvantizujeme s ditheringem na 8 bit?. 

173
00:09:09.086 --> 00:09:14.174
Na 2. kan?lu osciloskopu vid?me hladkou sinusovku.

174
00:09:14.174 --> 00:09:18.014
Zbl?zka vid?me o trochu v?c ?umu.

175
00:09:18.014 --> 00:09:19.305
To je vod?tko.

176
00:09:19.305 --> 00:09:21.273
Zkontrolujeme-li spektrum sign?lu...

177
00:09:22.889 --> 00:09:23.732
...aha!

178
00:09:23.732 --> 00:09:26.398
Na?e sinusovka je nedot?en?,

179
00:09:26.398 --> 00:09:28.490
ale ?rove? ?umu 8 bitov?ho sign?lu

180
00:09:28.490 --> 00:09:32.470
je mnohem vy???!

181
00:09:32.948 --> 00:09:36.148
A to je rozd?l, kter? d?l? po?et bit?.

182
00:09:36.148 --> 00:09:37.434
Nic v?c!

183
00:09:37.822 --> 00:09:39.956
Kdy? digitalizujeme sign?l,
nejd??v ho vzorkujeme.

184
00:09:39.956 --> 00:09:42.366
Vzorkovac? krok je p?esn?,
nevznik? ??dn? ztr?ta.

185
00:09:42.366 --> 00:09:45.626
Pak ale sign?l kvantizujeme
a kvantizace p?id?v? ?um.

186
00:09:47.827 --> 00:09:50.793
Po?et bit? ur?uje, kolik ?umu,

187
00:09:50.793 --> 00:09:52.569
tedy hladinu ?umu.

188
00:10:00.170 --> 00:10:03.646
Jak zn? kvantiza?n? ?um s ditheringem?

189
00:10:03.646 --> 00:10:06.012
Poslechneme si na?i 8-bitovou sinusovku.

190
00:10:12.521 --> 00:10:15.273
T??ko usly??me n?co jin?ho ne? jej? t?n.

191
00:10:15.273 --> 00:10:18.740
Odfiltrujeme t?n a poslechneme si samotn? ?um.

192
00:10:18.740 --> 00:10:21.683
A proto?e je ?um tich?, zv???me hlasitost.

193
00:10:32.009 --> 00:10:35.049
Pokud jste nahr?vali na analogov? za??zen?,

194
00:10:35.049 --> 00:10:36.670
mo?n? jste si pr?v? pomysleli:

195
00:10:36.670 --> 00:10:40.382
?No ne! To zn? jako ?um magnetick? p?sky!?

196
00:10:40.382 --> 00:10:41.929
Nejen ?e zn?,

197
00:10:41.929 --> 00:10:43.433
stejn? se i chov?.

198
00:10:43.433 --> 00:10:45.225
A pokud pou?ijeme Gauss?v dithering,

199
00:10:45.225 --> 00:10:47.646
je tento ?um a ?um p?sky matematicky shodn?.

200
00:10:47.646 --> 00:10:49.225
<u>Je</u> to ?um magnetick? p?sky.
 
201
00:10:49.225 --> 00:10:51.774
To znamen?, ?e m??eme m??it ?um p?sky

202
00:10:51.774 --> 00:10:54.196
a t?m p?dem ?rove? ?umu magnetick? p?sky

203
00:10:54.196 --> 00:10:56.233
pomoc? bit? m?sto decibel?.

204
00:10:56.233 --> 00:10:59.902
A pohl??et na n?j digit?ln?.

205
00:10:59.902 --> 00:11:03.028
Magnetofonov? kazety...

206
00:11:03.028 --> 00:11:05.449
Pokud je pamatujete,

207
00:11:05.449 --> 00:11:09.161
v nejlep??m p??pad? dos?hly kvality 9 bit?,

208
00:11:09.161 --> 00:11:11.209
v praxi sp??e 5 nebo 6,

209
00:11:11.209 --> 00:11:13.876
pokud jste nahr?vali pomoc? kaze??ku.

210
00:11:13.876 --> 00:11:19.422
Va?e nahr?vky m?ly hloubku 6 bit?... v lep??m p??pad?. 

211
00:11:19.837 --> 00:11:22.345
Profesion?ln? kotou?ov? p?sky

212
00:11:22.345 --> 00:11:24.553
pou??van? ve studi?ch st??? dos?hly...

213
00:11:24.553 --> 00:11:26.473
uh?dnete?...

214
00:11:26.473 --> 00:11:27.604
13 bit?.

215
00:11:27.604 --> 00:11:28.980
Pomoc? siln?ho potla?en? ?umu.

216
00:11:28.980 --> 00:11:32.062
Proto zkratka 'DDD' na kompaktn?ch disc?ch

217
00:11:32.062 --> 00:11:35.208
znamenala velk? high-end.

218
00:11:40.116 --> 00:11:42.825
St?le ??k?m, ?e kvantizuji s ditheringem.

219
00:11:42.825 --> 00:11:44.734
Co ale dithering znamen??

220
00:11:44.734 --> 00:11:47.284
A co d?l??

221
00:11:47.284 --> 00:11:49.876
Nejjednodu??? zp?sob, jak kvantizovat sign?l,

222
00:11:49.876 --> 00:11:52.329
je vybrat ??slicovou hodnotu

223
00:11:52.329 --> 00:11:54.377
nejbli??? okam?it? v?chylce.

224
00:11:54.377 --> 00:11:55.337
O?ividn?...

225
00:11:55.337 --> 00:11:57.545
Nane?t?st? konkr?tn? ?um,

226
00:11:57.545 --> 00:11:59.220
kter? vznikne t?mto postupem,

227
00:11:59.220 --> 00:12:02.174
je z?visl? na vstupn?m sign?lu.

228
00:12:02.174 --> 00:12:04.596
Tak?e vznikne ?um, kter? je nekonzistentn?,

229
00:12:04.596 --> 00:12:06.142
zp?sobuje zkreslen?,

230
00:12:06.142 --> 00:12:09.054
p??padn? je jinak ne??douc?.

231
00:12:09.054 --> 00:12:11.764
Dithering je speci?ln? zkonstruovan? ?um,

232
00:12:11.764 --> 00:12:15.273
kter? nahrazuje ?um vznikl? jednoduchou kvantizac?.

233
00:12:15.273 --> 00:12:18.025
Dithering nep?ehlu?uje
ani nemaskuje kvantiza?n? ?um,

234
00:12:18.025 --> 00:12:20.190
ale nahrazuje ho ?umem,

235
00:12:20.190 --> 00:12:22.612
kter? m? n?mi zvolen? charakteristiky,

236
00:12:22.612 --> 00:12:24.794
nez?visl? ne vstupn?m sign?lu.

237
00:12:25.256 --> 00:12:27.081
Pod?v?me se, co dithering d?l?.

238
00:12:27.081 --> 00:12:30.078
Proto?e gener?tor produkuje p??li? mnoho ?umu,

239
00:12:30.431 --> 00:12:33.161
budeme pro tento pokus generovat sinusovku

240
00:12:33.161 --> 00:12:34.782
pomoc? ThinkPadu.

241
00:12:34.782 --> 00:12:38.205
Kvantizujeme j? na 8 bit? s ditheringem.

242
00:12:39.006 --> 00:12:41.342
Na displeji vid?me p?knou sinusovku,

243
00:12:41.342 --> 00:12:43.452
stejn? jako na osciloskopu...

244
00:12:44.222 --> 00:12:44.972
a...

245
00:12:46.588 --> 00:12:49.375
a? se obnov? spektr?ln? analyz?tor...

246
00:12:50.713 --> 00:12:53.588
?istou frekven?n? ?pi?ku a konstantn? hladinu ?umu 

247
00:12:56.864 --> 00:12:58.611
na obou analyz?torech,

248
00:12:58.611 --> 00:12:59.646
stejn? jako p?edt?m.

249
00:12:59.646 --> 00:13:01.549
Opakuji, te? pou??v?m dithering.

250
00:13:02.196 --> 00:13:04.225
Te? dithering vypnu.

251
00:13:05.779 --> 00:13:07.913
Kvantiza?n? ?um, ditheringem rozprost?en?

252
00:13:07.913 --> 00:13:09.577
do konstantn? hladiny,

253
00:13:09.577 --> 00:13:12.286
vyroste do ?pi?ek harmonick?ho zkreslen?.

254
00:13:12.286 --> 00:13:16.030
Hladina ?umu je ni???, ale ?rove? zkreslen? je nenulov?

255
00:13:16.030 --> 00:13:19.668
a ?pi?ky zkreslen? jsou v??, ne? byla hladina ?umu p?i ditheringu.

256
00:13:19.668 --> 00:13:22.318
V p??pad? 8-bitov? hloubky je tento efekt p?ece?ov?n.

257
00:13:22.488 --> 00:13:24.200
P?i 16 bitech,

258
00:13:24.692 --> 00:13:25.929
dokonce i bez ditheringu,

259
00:13:25.929 --> 00:13:28.308
bude harmonick? zkreslen? velmi n?zk?,

260
00:13:28.308 --> 00:13:30.708
t?m?? nesly?iteln?.

261
00:13:30.708 --> 00:13:34.581
P?esto m??eme dithering pou??t

262
00:13:34.581 --> 00:13:36.489
pro ?plnou eliminaci zkreslen?.

263
00:13:37.642 --> 00:13:39.273
Op?t na chv?li dithering vypnu.

264
00:13:40.934 --> 00:13:43.444
V?imn?te si, ?e ?rove? zkreslen? 

265
00:13:43.444 --> 00:13:47.070
sign?lu kvantizovan?ho bez ditheringu

266
00:13:47.070 --> 00:13:49.033
je nez?visl? na amplitud? sign?lu.

267
00:13:49.033 --> 00:13:51.998
Kdy? ale amplituda klesne pod 1/2 bitu,

268
00:13:51.998 --> 00:13:54.036
v?e se kvantizuje na nulu.

269
00:13:54.036 --> 00:13:54.910
D? se ??ct,

270
00:13:54.910 --> 00:13:58.557
?e cokoliv kvantizov?no na nulu m? 100% zkreslen?!

271
00:13:58.833 --> 00:14:01.588
I toto zkreslen? dithering eliminuje.

272
00:14:01.588 --> 00:14:03.599
Zapneme dithering a...

273
00:14:03.599 --> 00:14:06.377
n?? sign?l je zp?t s ?rovn? 1/4 bitu,

274
00:14:06.377 --> 00:14:09.076
v?etn? vyrovnan? hladiny ?umu.

275
00:14:09.630 --> 00:14:11.220
Hladina ?umu nemus? b?t vyrovnan?.

276
00:14:11.220 --> 00:14:12.798
Dither je sign?l dle na?eho v?b?ru.

277
00:14:12.798 --> 00:14:15.006
Vyberme ?um tak nen?padn? a nepost?ehnuteln?,

278
00:14:15.006 --> 00:14:17.017
jak je jen mo?n?.

279
00:14:18.142 --> 00:14:22.484
N?? sluch je nejcitliv?j?? na p?smo 2kHz -- 4kHz,

280
00:14:22.484 --> 00:14:25.438
tak?e tam bude ?um nejv?c z?eteln?.

281
00:14:25.438 --> 00:14:29.406
M??eme tvarovat ditheringov? ?um

282
00:14:29.406 --> 00:14:31.241
a p?esunout ho do m?n? vn?man?ch oblast?,

283
00:14:31.241 --> 00:14:33.910
oby?ejn? k vy???m frekvenc?m.

284
00:14:34.249 --> 00:14:37.460
16-bitov? ditheringov? ?um je norm?ln? nesly?iteln?,

285
00:14:37.460 --> 00:14:39.668
ale poslechn?me si n?? tvarovan? ?um

286
00:14:39.668 --> 00:14:42.234
p?i mnohem vy??? hlasitosti...

287
00:14:56.020 --> 00:14:59.977
Kvantiza?n? ?um s ditheringem
p?en??? vy??? v?kon

288
00:14:59.977 --> 00:15:04.276
ne? bez ditheringu, p?esto?e zn? ti?eji.

289
00:15:04.276 --> 00:15:07.902
To je mo?n? vid?t na VU metru b?hem tich?ch pas???.

290
00:15:07.902 --> 00:15:10.537
Ale dithering m??eme
nejen zapnout nebo vypnout.

291
00:15:10.537 --> 00:15:14.712
M??eme sn??it jeho ?rove?
za cenu lehk?ho zkreslen?

292
00:15:14.712 --> 00:15:18.313
a minimalizovat tak v?sledn? efekt.

293
00:15:19.605 --> 00:15:22.790
Te? budeme vstupn? sign?l takto modulovat...

294
00:15:27.098 --> 00:15:30.206
...abychom uk?zali, jak to ovlivn? kvantiza?n? ?um.

295
00:15:30.206 --> 00:15:33.289
P?i pln? ?rovni je ?um ditheringu (podle p?edpokladu)

296
00:15:33.289 --> 00:15:35.643
rovnom?rn?, konstantn? a jednotv?rn?.

297
00:15:40.937 --> 00:15:42.772
Kdy? v?ak sni?ujeme jeho ?rove?,

298
00:15:42.772 --> 00:15:46.356
vstupn? sign?l st?le v?ce
ovliv?uje amplitudu a charakter

299
00:15:46.356 --> 00:15:47.977
kvantiza?n?ho ?umu.

300
00:16:09.883 --> 00:16:13.844
Tvarovan? dithering se chov? podobn?,

301
00:16:13.844 --> 00:16:16.553
ale tvarov?n? n?m poskytuje jednu v?hodu:

302
00:16:16.553 --> 00:16:18.804
M??eme pou??t dithering s ni??? ?rovn?

303
00:16:18.804 --> 00:16:20.937
p?i stejn?m ovlivn?n? v?stupn?ho sign?lu

304
00:16:20.937 --> 00:16:23.662
sign?lem vstupn?m.

305
00:16:49.172 --> 00:16:51.508
P?esto?e jsem str?vil ditheringem tolik ?asu,

306
00:16:51.508 --> 00:16:53.012
jedn? se o rozd?ly,

307
00:16:53.012 --> 00:16:56.372
kter? za??naj? na ?rovni -100 decibel?.

308
00:16:56.372 --> 00:16:59.806
Mo?n? by byl dithering d?le?it?j??,

309
00:16:59.806 --> 00:17:01.513
kdyby m?ly CD disky 14 bit?.

310
00:17:01.989 --> 00:17:02.644
Mo?n?.

311
00:17:02.644 --> 00:17:05.438
P?i 16 bitech opravdu nehraje roli.

312
00:17:05.438 --> 00:17:08.019
Dithering m??ete vn?mat jako pojistku,

313
00:17:08.019 --> 00:17:11.443
kter? n?m zajist? n?kolik
decibel? dynamick?ho rozsahu nav?c,

314
00:17:11.443 --> 00:17:12.804
jen pro p??pad.

315
00:17:12.990 --> 00:17:14.196
Pravdou je,

316
00:17:14.196 --> 00:17:16.361
?e vynech?n?m ditheringu

317
00:17:16.361 --> 00:17:19.182
dobrou nahr?vku nikdy nezni??te.

318
00:17:24.414 --> 00:17:25.790
Pou??vali jsme sinusovky.

319
00:17:25.790 --> 00:17:28.254
Jsou jasnou volbou,

320
00:17:28.254 --> 00:17:32.212
pokud chceme vid?t chov?n?
sign?lu izolovan? frekvence.

321
00:17:32.212 --> 00:17:34.217
Te? se pod?v?me na n?co slo?it?j??ho.

322
00:17:34.217 --> 00:17:35.923
Co bychom m?li o?ek?vat,

323
00:17:35.923 --> 00:17:39.671
pokud zm?n?m pr?b?h na obd?ln?kov??

324
00:17:42.718 --> 00:17:45.921
Vstupn? osciloskop ukazuje 1kHz obd?ln?ky.

325
00:17:45.921 --> 00:17:47.351
V?stupn? osciloskop ukazuje...

326
00:17:48.614 --> 00:17:51.102
p?esn? to, co by m?l.

327
00:17:51.102 --> 00:17:53.900
Co je vlastn? obd?ln?kov? sign?l?

328
00:17:54.654 --> 00:17:57.982
Je to sign?l s jistou kladnou hodnotou,

329
00:17:57.982 --> 00:18:00.788
kter? v polovin? periody skokem p?ejde

330
00:18:00.788 --> 00:18:02.910
do z?porn? hodnoty.

331
00:18:02.910 --> 00:18:05.076
T?m ov?em nevysv?tl?me,

332
00:18:05.076 --> 00:18:07.241
jak se z tohoto vstupu

333
00:18:07.241 --> 00:18:09.378
stal tento v?stup.

334
00:18:10.132 --> 00:18:12.713
Vzpome?me si, ?e ka?d? sign?l

335
00:18:12.713 --> 00:18:15.508
je sou?tem diskr?tn?ch frekvenc?

336
00:18:15.508 --> 00:18:18.302
a obd?ln?kov? sign?l je sou?tem

337
00:18:18.302 --> 00:18:19.636
z?kladn? frekvence

338
00:18:19.636 --> 00:18:22.228
a nekone?n?ho mno?stv? lich?ch harmonick?ch.

339
00:18:22.228 --> 00:18:24.597
Jejich sou?tem z?sk?me obd?ln?k.

340
00:18:26.398 --> 00:18:27.433
Na prvn? pohled

341
00:18:27.433 --> 00:18:29.225
n?m ani toto nepom??e.

342
00:18:29.225 --> 00:18:31.561
Mus?te se??st nekone?n? mno?stv? harmonick?ch frekvenc?,

343
00:18:31.561 --> 00:18:33.108
abyste dostali odpov??.

344
00:18:33.108 --> 00:18:35.977
My ale nem?me
nekone?n? mno?stv? frekvenc?!

345
00:18:36.960 --> 00:18:39.902
Pou??v?me ostr? antialiasingov? filtr,

346
00:18:39.902 --> 00:18:42.206
kter? omez? sign?l t?sn? nad 20kHz,

347
00:18:42.206 --> 00:18:44.158
tak?e n?? sign?l je frekven?n? omezen

348
00:18:44.158 --> 00:18:46.421
a m?me toto:

349
00:18:52.500 --> 00:18:56.468
...a p?esn? to vid?me na osciloskopu.

350
00:18:56.468 --> 00:18:59.550
Zvln?n? okolo ostr?ch hran omezen?ho sign?lu

351
00:18:59.550 --> 00:19:00.926
naz?v?me Gibbs?v efekt.

352
00:19:00.926 --> 00:19:04.137
Projev? se poka?d?, kdy? odstran?te
oblast frekven?n?ho spektra

353
00:19:04.137 --> 00:19:07.006
v oblasti nenulov? energie.

354
00:19:07.006 --> 00:19:09.854
Jednoduch? pravidlo zn?: ??m ost?ej?? omezen?,

355
00:19:09.854 --> 00:19:11.188
t?m v?t?? zvln?n?.

356
00:19:11.188 --> 00:19:12.777
Co? je zhruba pravda,

357
00:19:12.777 --> 00:19:14.900
ale je pot?eba b?t opatrn?.

358
00:19:14.900 --> 00:19:15.774
Nap??klad...

359
00:19:15.774 --> 00:19:19.529
Co o?ek?v?te, ?e se stane, kdy? n?? sign?l projde

360
00:19:19.529 --> 00:19:23.181
na??m antialiasingov?m filtrem podruh??

361
00:19:34.136 --> 00:19:37.588
Krom zpo?d?n? o n?kolik cykl?...

362
00:19:37.588 --> 00:19:39.348
je odpov??...

363
00:19:39.348 --> 00:19:40.857
v?bec nic.

364
00:19:41.257 --> 00:19:43.302
Sign?l u? byl frekven?n? omezen.

365
00:19:43.656 --> 00:19:46.590
Nov? omezen? s n?m neud?l? nic.

366
00:19:46.590 --> 00:19:50.686
Druh? pr?chod neodstran? frekvence,
kter? u? byly jednou odstran?ny.

367
00:19:52.070 --> 00:19:53.737
To je d?le?it?.

368
00:19:53.737 --> 00:19:56.233
Lid? ?asto pova?uj? zvln?n? za artefakt,

369
00:19:56.233 --> 00:19:59.945
kter? je p?id?n antialiasingov?m
nebo rekonstruk?n?m filtrem,

370
00:19:59.945 --> 00:20:01.737
a ?e se zvln?n?

371
00:20:01.737 --> 00:20:03.913
ka?d?m pr?chodem zhor??.

372
00:20:03.913 --> 00:20:05.950
Zde vid?me, ?e to se nestalo.

373
00:20:05.950 --> 00:20:09.492
Byl to tedy filtr,
kter? p?i prvn?m pr?chodu p?idal zvln?n??

374
00:20:09.492 --> 00:20:10.537
Rozhodn? ne.

375
00:20:10.537 --> 00:20:12.126
Je to jemn? nuance,

376
00:20:12.126 --> 00:20:15.252
ale zvln?n? od Gibbsova efektu nen? p?id?no filtrem,

377
00:20:15.252 --> 00:20:18.836
ale je prost? sou??st? frekven?n? omezen?ho sign?lu.

378
00:20:18.836 --> 00:20:20.798
I kdy? synteticky vyrob?me sign?l,

379
00:20:20.798 --> 00:20:23.508
kter? vypad? jako dokonal? obd?ln?k,

380
00:20:23.508 --> 00:20:26.206
st?le je omezen? ???kou p?sma.

381
00:20:26.206 --> 00:20:29.140
Vzpome?te si, ?e schodov? graf je zav?d?j?c?.

382
00:20:29.140 --> 00:20:32.222
Ve skute?nosti m?me
jen body v okam?ic?ch vzorkov?n? 

383
00:20:32.222 --> 00:20:36.148
a pr?v? jeden frekven?n?
omezen? sign?l jim vyhovuje.

384
00:20:36.148 --> 00:20:39.614
Kdy? jsme nakreslili dokonal? obd?ln?kov? sign?l,

385
00:20:39.614 --> 00:20:43.198
jen jsme pospojovali body jako p?i spojova?ce,

386
00:20:43.198 --> 00:20:47.785
aby to vypadalo, ?e nem? ??dn? zvln?n?.

387
00:20:47.785 --> 00:20:49.449
Ale p?vodn?, frekven?n? omezen? sign?l

388
00:20:49.449 --> 00:20:52.742
v?etn? zvln?n? je po??d zde.

389
00:20:54.004 --> 00:20:56.542
To n?s p?iv?d? k d?le?it?mu bodu.

390
00:20:56.542 --> 00:20:59.550
Nejsp?? jste sly?eli, ?e p?esnost ?asov?n? 

391
00:20:59.550 --> 00:21:02.409
je limitov?na vzorkovac?
frekvenc? digit?ln?ho sign?lu.

392
00:21:02.409 --> 00:21:05.140
Jinak ?e?eno: digit?ln? sign?l
nem??e zachytit cokoliv,

393
00:21:05.140 --> 00:21:08.041
co se odehraje mezi vzorky...

394
00:21:08.041 --> 00:21:11.422
z toho by plynulo,
?e strm? hrany mus? b?t zarovn?ny

395
00:21:11.422 --> 00:21:14.473
p?esn? se vzorkov?n?m,
jinak bude ?asov?n? posunuto,

396
00:21:14.473 --> 00:21:16.219
nebo hrany prost? zmiz?.

397
00:21:16.711 --> 00:21:20.820
U? te? vid?me, pro? to nen? pravda.

398
00:21:20.820 --> 00:21:23.742
Opakuji, na?e sign?ly jsou frekven?n? omezen?.

399
00:21:23.742 --> 00:21:26.036
A digit?ln? sign?l, to jsou vzorky,

400
00:21:26.036 --> 00:21:29.340
ne schody nebo ?spojova?ka?.

401
00:21:31.572 --> 00:21:34.592
Rozhodn? nap??klad m??eme...

402
00:21:36.777 --> 00:21:39.337
um?stit n?b??nou hranu
omezen?ho obd?ln?kov?ho sign?lu

403
00:21:39.337 --> 00:21:42.004
kamkoliv mezi vzorky.

404
00:21:42.004 --> 00:21:44.354
Je dokonale pops?na...

405
00:21:47.508 --> 00:21:50.218
a dokonale zrekonstruov?na.

406
00:22:04.620 --> 00:22:06.526
Stejn? jako v p?edchoz?m d?le

407
00:22:06.526 --> 00:22:08.393
jsme probrali mnoho t?mat,

408
00:22:08.393 --> 00:22:10.868
av?ak zdaleka ne do hloubky.

409
00:22:10.868 --> 00:22:13.620
Pokud nic jin?ho, m? h??chy
opominut? se t?mto prohloubily...

410
00:22:13.620 --> 00:22:16.286
T?mto bychom mohli skon?it.

411
00:22:16.286 --> 00:22:17.833
Anebo za??t?

412
00:22:17.833 --> 00:22:18.708
Zkoumejte do hloubky.

413
00:22:18.708 --> 00:22:19.710
Experimentujte.

414
00:22:19.710 --> 00:22:21.374
Svoje pokusy jsem vybral pe?liv?,

415
00:22:21.374 --> 00:22:23.668
aby byly jednoduch? a s jasn?mi v?sledky.

416
00:22:23.668 --> 00:22:26.217
Kter?koliv z nich si m??ete sami zopakovat.

417
00:22:26.217 --> 00:22:28.766
P?esto se n?kdy nau??me nejv?c t?m,

418
00:22:28.766 --> 00:22:30.516
?e rozbijeme obl?benou hra?ku

419
00:22:30.516 --> 00:22:32.553
a prozkoum?me d?ly, kter? z n? vypadly.

420
00:22:32.553 --> 00:22:35.230
To je v po??dku, jsme in?en??i.

421
00:22:35.230 --> 00:22:36.350
Hrajte si s parametry,

422
00:22:36.350 --> 00:22:37.972
hackujte k?d,

423
00:22:37.972 --> 00:22:39.774
vym??lejte podobn? pokusy.

424
00:22:39.774 --> 00:22:40.692
Zdrojov? k?dy ke v?emu,

425
00:22:40.692 --> 00:22:42.398
v?etn? p?edv?d?c? aplikace s tla??tky,

426
00:22:42.398 --> 00:22:44.361
je k nalezen? na Xiph.Org.

427
00:22:44.361 --> 00:22:45.940
P?i experimentov?n?

428
00:22:45.940 --> 00:22:47.401
pravd?podobn? naraz?te na n?co neo?ek?van?ho,

429
00:22:47.401 --> 00:22:49.950
?emu nebudete rozum?t.

430
00:22:49.950 --> 00:22:51.198
??dn? strach!

431
00:22:51.198 --> 00:22:54.537
Nehled? na moj? p?edchoz? pozn?mku o Wikipedii,

432
00:22:54.537 --> 00:22:56.788
je v?born?m m?stem pro p??le?itostn? v?zkum.

433
00:22:56.788 --> 00:22:59.956
A pokud to mysl?te se zpracov?n?m sign?l? v??n?,

434
00:22:59.956 --> 00:23:03.337
r?zn? vysok? ?koly poskytuj? pokro?il? materi?ly,

435
00:23:03.337 --> 00:23:07.380
nap??klad kurzy 6.003 a 6.007 -- ?Signals and Systems?

436
00:23:07.380 --> 00:23:08.798
na MIT OpenCourseWare.

437
00:23:08.798 --> 00:23:11.593
A samoz?ejm? je zde na?e komunita na Xiph.Org.

438
00:23:12.792 --> 00:23:13.929
Pokusy nepokusy,

439
00:23:13.929 --> 00:23:14.974
do?la mi k?va.

440
00:23:14.974 --> 00:23:16.436
Tak?e nashledanou.

441
00:23:16.436 --> 00:23:19.316
happy hacking!