1
00:00:04,160 --> 00:00:07,200
Speaker 1: Get in touch with technology with tech Stuff from half

2
00:00:07,240 --> 00:00:14,720
Speaker 1: stuff works dot com. Hey everybody, it's Jonathan Strickland here

3
00:00:14,920 --> 00:00:19,120
Speaker 1: with text Stuff classic episodes. We're doing some Saturday morning

4
00:00:19,239 --> 00:00:22,439
Speaker 1: reruns for you guys. This is a special series where

5
00:00:22,440 --> 00:00:25,440
Speaker 1: we're going to dig up some classic episodes of tech

6
00:00:25,520 --> 00:00:28,400
Speaker 1: Stuff and present them to you guys who may not

7
00:00:28,520 --> 00:00:30,480
Speaker 1: have had a chance to listen to them, especially if

8
00:00:30,480 --> 00:00:33,400
Speaker 1: you're a brand new listener. First of all, welcome. If

9
00:00:33,440 --> 00:00:36,480
Speaker 1: that's the case, I hope you enjoy these episodes. This

10
00:00:36,520 --> 00:00:39,800
Speaker 1: one is called The Dirt on Digital Audio, and it

11
00:00:39,880 --> 00:00:43,960
Speaker 1: is an episode all about the actual technical process of

12
00:00:44,040 --> 00:00:48,200
Speaker 1: recording audio into a digital format and what that requires

13
00:00:48,240 --> 00:00:51,680
Speaker 1: because it's very different from the analog style. I hope

14
00:00:51,720 --> 00:00:55,280
Speaker 1: you guys enjoy it. This episode was originally published on

15
00:00:55,360 --> 00:00:58,960
Speaker 1: November twenty three, two thousand sixteen. And just in case

16
00:00:58,960 --> 00:01:01,120
Speaker 1: you're listening to this one in the far future, I'm

17
00:01:01,160 --> 00:01:04,440
Speaker 1: recording this in two thousand eighteen, So we're gonna time

18
00:01:04,440 --> 00:01:08,000
Speaker 1: travel a bit and listen to this classic episode The

19
00:01:08,080 --> 00:01:12,080
Speaker 1: Dirt on Digital Audio. So to start it all off,

20
00:01:12,880 --> 00:01:15,840
Speaker 1: we all have to take a quick trip to Germany,

21
00:01:16,000 --> 00:01:19,520
Speaker 1: So anyone who is not in Germany get your passport.

22
00:01:20,240 --> 00:01:22,280
Speaker 1: I was actually in Germany not that long ago. I

23
00:01:22,319 --> 00:01:25,800
Speaker 1: got to visit Berlin and had a wonderful time. And

24
00:01:25,880 --> 00:01:29,279
Speaker 1: in Germany there's a company called frown Hofer Gazelle Shaft.

25
00:01:30,080 --> 00:01:32,000
Speaker 1: And you might wonder, well, what does this company do

26
00:01:32,440 --> 00:01:38,040
Speaker 1: they think? I joke that my profession, that my title

27
00:01:38,200 --> 00:01:40,800
Speaker 1: that I should put on my business card it should

28
00:01:40,800 --> 00:01:44,399
Speaker 1: say professional smart person. Well, no joke, that's what these

29
00:01:44,400 --> 00:01:49,920
Speaker 1: people are. They specialize in research and development, applied research.

30
00:01:51,080 --> 00:01:54,520
Speaker 1: It's a whole company that specializes and applied research. And

31
00:01:54,640 --> 00:01:58,960
Speaker 1: it's huge. It encompasses sixties seven institutes and research units

32
00:01:59,040 --> 00:02:04,480
Speaker 1: across Germany. Well back in the eighties and there was

33
00:02:04,520 --> 00:02:10,320
Speaker 1: a researcher named Karl Heinz Brandenburg, and Karl Heinz made

34
00:02:10,720 --> 00:02:16,720
Speaker 1: a breakthrough round uh and came up with this clever

35
00:02:16,880 --> 00:02:21,360
Speaker 1: idea about encoding audio. He was actually working towards creating

36
00:02:21,360 --> 00:02:25,640
Speaker 1: a way that would allow for high audio quality transfer

37
00:02:26,040 --> 00:02:30,239
Speaker 1: but having a low bit rate sampling, so that file

38
00:02:30,360 --> 00:02:34,160
Speaker 1: sizes and transfer times wouldn't get out of control. Because

39
00:02:34,160 --> 00:02:36,040
Speaker 1: you got to remember, this is the eighties, this is

40
00:02:36,120 --> 00:02:39,200
Speaker 1: before the World Wide Web was a thing that would

41
00:02:39,360 --> 00:02:42,400
Speaker 1: that wouldn't happen until the early nineties, so the Internet

42
00:02:42,440 --> 00:02:44,280
Speaker 1: was very young. In fact, they weren't even looking at

43
00:02:44,320 --> 00:02:47,320
Speaker 1: the Internet as a method of distribution for this particular

44
00:02:47,400 --> 00:02:50,639
Speaker 1: type of encoded audio. They were looking at using this

45
00:02:51,000 --> 00:02:54,880
Speaker 1: to transmit across telephone lines. So they needed to have

46
00:02:54,960 --> 00:02:58,320
Speaker 1: something that was going to be high quality but low space.

47
00:02:59,520 --> 00:03:01,600
Speaker 1: So what the heck does that mean? All right? Well,

48
00:03:02,560 --> 00:03:06,960
Speaker 1: digital audio and analog audio are very different things. So

49
00:03:07,000 --> 00:03:10,320
Speaker 1: to understand that, we need to look at how sound

50
00:03:10,360 --> 00:03:14,679
Speaker 1: works and how we describe sound, because that informs how

51
00:03:14,720 --> 00:03:19,440
Speaker 1: we can capture sound and replicate those qualities digitally. So

52
00:03:19,520 --> 00:03:22,560
Speaker 1: stick with me. We're gonna go back to school for

53
00:03:22,720 --> 00:03:27,400
Speaker 1: some basic sound science. And this goes back to the

54
00:03:27,400 --> 00:03:31,720
Speaker 1: way sound physically moves through a medium, whether that's a

55
00:03:31,800 --> 00:03:36,640
Speaker 1: solid or through the air or through water. Sound is vibration.

56
00:03:37,480 --> 00:03:42,640
Speaker 1: Now we sense this primarily through hearing it or sometimes

57
00:03:42,880 --> 00:03:45,720
Speaker 1: feeling it. If it's the right frequency and the right amplitude.

58
00:03:45,720 --> 00:03:49,240
Speaker 1: We can actually feel sound. Anyone who stood close to,

59
00:03:49,280 --> 00:03:52,320
Speaker 1: say a sub wiffer that was really blasting out bass notes,

60
00:03:52,400 --> 00:03:54,600
Speaker 1: you know what I'm talking about. You can feel it

61
00:03:54,720 --> 00:03:58,720
Speaker 1: pressing against you. Well, sound travels through the air when

62
00:03:58,760 --> 00:04:03,320
Speaker 1: molecules vibrate against each other, and this creates instances of

63
00:04:03,560 --> 00:04:07,840
Speaker 1: increased pressure and decreased pressure at what is a hyper

64
00:04:08,000 --> 00:04:11,000
Speaker 1: local level. We're not talking about weather maps here, We're

65
00:04:11,040 --> 00:04:15,160
Speaker 1: talking about tiny little areas. So this increase in decrease

66
00:04:15,160 --> 00:04:17,760
Speaker 1: in pressure is something that we can sense as sound.

67
00:04:18,320 --> 00:04:21,520
Speaker 1: When those changes in pressure affect a diaphragm, such as

68
00:04:21,640 --> 00:04:25,640
Speaker 1: one that's in a microphone or maybe your ear drum,

69
00:04:25,839 --> 00:04:30,040
Speaker 1: for example, it causes the diaphragm to actually move. So

70
00:04:30,200 --> 00:04:36,839
Speaker 1: increased pressure pushes the diaphragm in and decreased pressure doesn't

71
00:04:36,920 --> 00:04:39,440
Speaker 1: really pull the diaphragm out. I mean, you could say

72
00:04:39,440 --> 00:04:42,080
Speaker 1: it it pulls the diaphragm out, but to be more accurate,

73
00:04:42,120 --> 00:04:46,080
Speaker 1: the diaphragm actually pushes outward because the pressure on the

74
00:04:46,080 --> 00:04:48,640
Speaker 1: outside is lower than the pressure on the inside. But

75
00:04:48,680 --> 00:04:51,839
Speaker 1: you get what I'm saying. The diaphragm begins to to

76
00:04:52,440 --> 00:04:56,200
Speaker 1: flex inward and outward depending upon the amount of pressure

77
00:04:56,279 --> 00:04:59,720
Speaker 1: that it's it's encountering. You can imagine this being kind

78
00:04:59,720 --> 00:05:01,919
Speaker 1: of like a drum drum, not an ear drum, but

79
00:05:01,920 --> 00:05:04,800
Speaker 1: an actual drum and striking it. Uh, that's the same

80
00:05:04,839 --> 00:05:08,920
Speaker 1: sort of thing. So sound is the fluctuations of pressure,

81
00:05:09,480 --> 00:05:12,760
Speaker 1: which we can diagram as a wave or a wave

82
00:05:12,839 --> 00:05:16,440
Speaker 1: length a wave form on an x y axis, So

83
00:05:16,480 --> 00:05:21,560
Speaker 1: the horizontal line that access that represents time that has passed,

84
00:05:21,960 --> 00:05:26,000
Speaker 1: and the vertical axis represents the amplitude or the volume

85
00:05:26,560 --> 00:05:29,880
Speaker 1: of the sound wave. The wave length of the sound

86
00:05:30,240 --> 00:05:32,760
Speaker 1: which is the distance between successive points on a wave,

87
00:05:32,839 --> 00:05:36,359
Speaker 1: such as like the successive crests on a wave. That

88
00:05:36,440 --> 00:05:39,600
Speaker 1: tells you a lot about the frequency. So sound moves

89
00:05:39,640 --> 00:05:43,280
Speaker 1: at a constant rate through a given medium, but it

90
00:05:43,320 --> 00:05:47,200
Speaker 1: moves at different rates through different media. So in other words,

91
00:05:47,440 --> 00:05:49,760
Speaker 1: it moves a different speed through a solid than it

92
00:05:49,800 --> 00:05:53,400
Speaker 1: does through air. If the crests of each sound wave

93
00:05:53,480 --> 00:05:57,479
Speaker 1: are really close together, that's a high frequency sound. More

94
00:05:57,520 --> 00:06:00,920
Speaker 1: waves will pass through an arbitrary point within a second.

95
00:06:01,360 --> 00:06:04,240
Speaker 1: The waves that are spaced further apart, that would be

96
00:06:04,279 --> 00:06:07,559
Speaker 1: a lower frequency sound. Higher frequency sounds have a higher

97
00:06:07,600 --> 00:06:10,719
Speaker 1: pitch than lower frequency sounds. So if you hold a

98
00:06:10,800 --> 00:06:14,360
Speaker 1: single note at a constant frequency, you'll have what is

99
00:06:14,400 --> 00:06:18,520
Speaker 1: called a simple harmonic motion. That means the vibrations are

100
00:06:18,600 --> 00:06:21,880
Speaker 1: moving at a constant rate inward and outward. The cycle

101
00:06:22,120 --> 00:06:25,680
Speaker 1: is constant. A tuning fork is a good example of this.

102
00:06:26,800 --> 00:06:31,080
Speaker 1: So if you hear a clear C note played on

103
00:06:31,120 --> 00:06:34,360
Speaker 1: a musical instrument, that could be a simple harmonic motion.

104
00:06:34,600 --> 00:06:36,720
Speaker 1: It won't be, but it could be. I'll tell you

105
00:06:36,720 --> 00:06:39,080
Speaker 1: why it won't be in a minute. So the frequency

106
00:06:39,120 --> 00:06:42,240
Speaker 1: of vibration doesn't change, and so you would get this

107
00:06:42,480 --> 00:06:44,840
Speaker 1: very clear note as a result, And if you were

108
00:06:44,839 --> 00:06:49,760
Speaker 1: to diagram it, you would have very regular crests and troughs,

109
00:06:49,800 --> 00:06:53,600
Speaker 1: all of the same amplitude and distance from each other.

110
00:06:53,640 --> 00:06:58,240
Speaker 1: The frequency and volume would remain constant, assuming of course,

111
00:06:58,320 --> 00:07:02,480
Speaker 1: that you're not trying to change the frequency or volume. Now,

112
00:07:02,520 --> 00:07:05,320
Speaker 1: this is where I point out most musical instruments don't

113
00:07:05,400 --> 00:07:09,920
Speaker 1: produce a single clear note, even if played expertly. They

114
00:07:09,920 --> 00:07:15,360
Speaker 1: actually create several resonant frequencies. So every physical object resonates

115
00:07:15,400 --> 00:07:19,400
Speaker 1: at several different frequencies. You've probably seen this in various programs.

116
00:07:19,440 --> 00:07:22,840
Speaker 1: MythBusters did one about bridges, the idea being that if

117
00:07:22,880 --> 00:07:25,840
Speaker 1: you were to have a group of people marching on

118
00:07:25,880 --> 00:07:28,960
Speaker 1: a bridge at the bridge's resonant frequency, it could cause

119
00:07:29,000 --> 00:07:33,600
Speaker 1: the bridge to start to vibrate and swing out of control. Well,

120
00:07:33,640 --> 00:07:35,480
Speaker 1: there's a reason for this. You may have also seen

121
00:07:35,560 --> 00:07:39,160
Speaker 1: videos of people singing a certain note and causing a

122
00:07:39,240 --> 00:07:43,640
Speaker 1: crystal glass to shatter. That's because that crystal glass does

123
00:07:43,680 --> 00:07:45,880
Speaker 1: have a resonant frequency, and if you can hit that

124
00:07:45,920 --> 00:07:49,200
Speaker 1: resonant frequency at the right volume, you can cause the

125
00:07:49,240 --> 00:07:52,360
Speaker 1: glass to start to deform, or the crystal in this case,

126
00:07:52,440 --> 00:07:55,160
Speaker 1: to deform to a point where it loses integrity and

127
00:07:55,160 --> 00:08:00,840
Speaker 1: it shatters as a result. Well, the resonation of an

128
00:08:00,840 --> 00:08:04,560
Speaker 1: object is dependent upon lots of different factors, and in fact,

129
00:08:04,720 --> 00:08:09,760
Speaker 1: most stuff will resonate at different frequencies, but at different intensities.

130
00:08:10,040 --> 00:08:14,239
Speaker 1: Like there might be one sweet spot, one specific frequency

131
00:08:14,320 --> 00:08:18,600
Speaker 1: that will have the greatest effect, but other related frequencies

132
00:08:18,640 --> 00:08:20,480
Speaker 1: may also have an effect. It will just be to

133
00:08:20,520 --> 00:08:24,040
Speaker 1: a lesser extent. Well, if you were to pluck a

134
00:08:24,040 --> 00:08:28,240
Speaker 1: guitar string, just you've tuned it to whatever note doesn't matter.

135
00:08:28,320 --> 00:08:31,640
Speaker 1: Let's say it's you tuned it to to G and

136
00:08:31,880 --> 00:08:35,920
Speaker 1: you play the G string on your guitar. The note

137
00:08:35,960 --> 00:08:38,960
Speaker 1: that you will hear really over all others will be

138
00:08:39,000 --> 00:08:40,640
Speaker 1: g that that is going to be the one that

139
00:08:40,640 --> 00:08:43,320
Speaker 1: will sound the loudest, But it will also play resonant

140
00:08:43,320 --> 00:08:47,240
Speaker 1: frequencies at a decreased amplitude. In other words, of decreased

141
00:08:47,360 --> 00:08:51,440
Speaker 1: volume so you still hear the intended note above everything else,

142
00:08:51,480 --> 00:08:54,600
Speaker 1: above all the other resonant frequencies. This is called a

143
00:08:54,679 --> 00:08:58,800
Speaker 1: complex tone, and that collection of frequencies in their amplitudes

144
00:08:59,000 --> 00:09:03,720
Speaker 1: is called the sectrum of sound. You get a full spectrum. Now,

145
00:09:03,760 --> 00:09:09,280
Speaker 1: some of the components of that complex tone will be uh,

146
00:09:09,320 --> 00:09:12,360
Speaker 1: imperceptible to you. You there'll be so quiet that you

147
00:09:12,400 --> 00:09:15,640
Speaker 1: wouldn't really notice them. They might affect the overall quality

148
00:09:15,640 --> 00:09:17,280
Speaker 1: of the sound, but in such a subtle way that

149
00:09:17,320 --> 00:09:19,160
Speaker 1: it may be difficult for you to even put it

150
00:09:19,240 --> 00:09:23,200
Speaker 1: into words. Each of those little components is called a partial.

151
00:09:23,640 --> 00:09:25,959
Speaker 1: So in the example of a guitar string, the partials

152
00:09:26,000 --> 00:09:30,080
Speaker 1: are all integers of the same fundamental frequency, and the

153
00:09:30,160 --> 00:09:34,680
Speaker 1: sound has a harmonic spectrum. But as you get further

154
00:09:34,760 --> 00:09:39,600
Speaker 1: away from that fundamental frequency, the amplitude decreases significantly. So,

155
00:09:39,679 --> 00:09:42,719
Speaker 1: like I said, you get far enough away, they are

156
00:09:42,760 --> 00:09:47,040
Speaker 1: technically there, but they might be imperceptible to you. Now,

157
00:09:47,080 --> 00:09:51,520
Speaker 1: some sounds have frequencies that aren't integers of a fundamental

158
00:09:51,559 --> 00:09:55,320
Speaker 1: frequency and are inharmonic Uh. Certain bells, like if you

159
00:09:55,320 --> 00:09:57,240
Speaker 1: hear a bell ring, you can probably pick out a

160
00:09:57,240 --> 00:10:00,840
Speaker 1: couple of different frequencies. There that are not harmon frequencies.

161
00:10:01,679 --> 00:10:04,439
Speaker 1: These are very complex sounds, and to our perception, if

162
00:10:04,480 --> 00:10:07,480
Speaker 1: it's complex enough, it can seem like there's no single

163
00:10:07,559 --> 00:10:12,480
Speaker 1: discernible pitch. They're like there's no fundamental frequency over all

164
00:10:12,559 --> 00:10:16,640
Speaker 1: the others. If it's complex enough, we call it noise.

165
00:10:17,360 --> 00:10:21,040
Speaker 1: That is the technical term. It is noise. Now, the

166
00:10:21,160 --> 00:10:26,319
Speaker 1: unit we use to measure frequency is the hurts uh

167
00:10:26,559 --> 00:10:29,839
Speaker 1: H E R t Z. Typical human hearing ranges from

168
00:10:29,880 --> 00:10:33,840
Speaker 1: twenty hurts, which means a wave will pass a given

169
00:10:33,960 --> 00:10:37,400
Speaker 1: arbitrary point twenty times within a second, all the way

170
00:10:37,480 --> 00:10:40,439
Speaker 1: up to twenty killer hurts, which means a wave will

171
00:10:40,440 --> 00:10:44,520
Speaker 1: pass a particular point in time twenty thousand times in

172
00:10:44,520 --> 00:10:47,320
Speaker 1: a second, or particular point on your wave form twenty

173
00:10:47,320 --> 00:10:50,880
Speaker 1: thousand times in the second. And most of our sensitivity

174
00:10:51,040 --> 00:10:54,800
Speaker 1: tends to be between one or two killer hurts up

175
00:10:54,840 --> 00:10:58,000
Speaker 1: to four or five killer hurts. That's generally where we

176
00:10:58,240 --> 00:11:02,280
Speaker 1: have human voices, and we've really gotten good at picking

177
00:11:02,280 --> 00:11:04,800
Speaker 1: those out of over everything else. So our sensitivity of

178
00:11:04,880 --> 00:11:07,520
Speaker 1: hearing is really concentrated between one killer hurts and four

179
00:11:07,600 --> 00:11:10,640
Speaker 1: killer hurts or two and five depending upon whom you ask.

180
00:11:12,840 --> 00:11:16,200
Speaker 1: Now we get back over to amplitude, that is referring

181
00:11:16,240 --> 00:11:18,520
Speaker 1: to the height of the wave. It also refers to

182
00:11:18,559 --> 00:11:23,679
Speaker 1: the volume the loudness of something. Amplitude means bigness. So

183
00:11:23,720 --> 00:11:27,199
Speaker 1: how big is the sound, Well, the greater the amplitude,

184
00:11:27,240 --> 00:11:30,319
Speaker 1: the louder it is. And amplitudes can have an enormous

185
00:11:30,480 --> 00:11:34,080
Speaker 1: range and affect how we perceive sounds. So, for example,

186
00:11:34,559 --> 00:11:38,720
Speaker 1: take a really complicated classical piece of music. It's just

187
00:11:38,840 --> 00:11:42,120
Speaker 1: easy to explain it in that term. You might have

188
00:11:42,160 --> 00:11:45,760
Speaker 1: a stretch in that classical piece of music in which

189
00:11:45,840 --> 00:11:48,360
Speaker 1: all the instruments are more or less playing at a

190
00:11:48,440 --> 00:11:52,000
Speaker 1: similar volume, so the sound from each instrument section has

191
00:11:52,040 --> 00:11:55,319
Speaker 1: a similar amplitude. But then there might be one segment

192
00:11:55,400 --> 00:11:58,920
Speaker 1: where an instrument group or maybe even a single soloist

193
00:11:59,559 --> 00:12:03,600
Speaker 1: has an increased amplitude and increased volume. It rises over

194
00:12:03,640 --> 00:12:06,960
Speaker 1: the rest of the orchestra, and that peak of the

195
00:12:07,000 --> 00:12:10,600
Speaker 1: amplitude is called the attack of the sound, and the

196
00:12:10,880 --> 00:12:16,040
Speaker 1: entire range of amplitudes is called the amplitude envelope. Now

197
00:12:16,040 --> 00:12:18,920
Speaker 1: this is important when we get to m P three's

198
00:12:18,960 --> 00:12:23,640
Speaker 1: because the way we perceive these sounds uh that that

199
00:12:23,679 --> 00:12:26,240
Speaker 1: has everything to do with the way the MP three

200
00:12:26,360 --> 00:12:29,600
Speaker 1: was designed. The whole point of the MP three was

201
00:12:29,640 --> 00:12:34,800
Speaker 1: to try and create a small file size to represent

202
00:12:34,880 --> 00:12:37,880
Speaker 1: what we can hear and kind of ignore everything else.

203
00:12:38,120 --> 00:12:40,760
Speaker 1: We'll get to that in a little bit more more time.

204
00:12:40,920 --> 00:12:43,880
Speaker 1: So this is really interesting to me. If you take

205
00:12:44,240 --> 00:12:49,679
Speaker 1: a sound and you double its amplitude, you increase the

206
00:12:49,720 --> 00:12:54,080
Speaker 1: amplitude by twofold, a listener would not necessarily feel that

207
00:12:54,120 --> 00:12:59,400
Speaker 1: the sound is twice as loud. Human hearing is incredibly subjective,

208
00:13:00,040 --> 00:13:04,079
Speaker 1: and typically for most listeners, it would require much more

209
00:13:04,880 --> 00:13:08,760
Speaker 1: than doubling the sounds amplitude for them to feel that

210
00:13:08,880 --> 00:13:12,400
Speaker 1: the sound itself was twice as loud. This perception of

211
00:13:12,480 --> 00:13:14,920
Speaker 1: volume is important when we get to the lossy formats

212
00:13:14,920 --> 00:13:19,839
Speaker 1: for audio files. Now I've given you all this information,

213
00:13:20,040 --> 00:13:22,760
Speaker 1: and I know everyone is probably thinking, you know, I

214
00:13:23,120 --> 00:13:26,480
Speaker 1: learned this in primary school, elementary school. All of this

215
00:13:26,559 --> 00:13:29,800
Speaker 1: is really familiar to me, and you're maybe rolling your

216
00:13:29,800 --> 00:13:32,840
Speaker 1: eyes because it's so basic. But I think it's important

217
00:13:33,280 --> 00:13:36,560
Speaker 1: to have that refresher so that you can understand the

218
00:13:36,600 --> 00:13:41,240
Speaker 1: difference between sound as we experience it and sound as

219
00:13:41,320 --> 00:13:45,960
Speaker 1: the way we encode it digitally and replicate it digitally.

220
00:13:46,840 --> 00:13:49,840
Speaker 1: For one thing, this illustrates how sound in the real

221
00:13:49,880 --> 00:13:54,640
Speaker 1: world is a continuum. It's a continuum both in frequency

222
00:13:54,679 --> 00:13:59,920
Speaker 1: and amplitude. You can have sound changing in frequency very

223
00:14:00,280 --> 00:14:04,520
Speaker 1: smoothly from one pitch to another. You can also have

224
00:14:04,600 --> 00:14:09,199
Speaker 1: sound increase or decrease in amplitude in a very smooth way.

225
00:14:09,360 --> 00:14:14,240
Speaker 1: And it is continuous, it's unbroken. It can have smooth transitions.

226
00:14:14,240 --> 00:14:17,240
Speaker 1: And these qualities provide challenges when we want to describe

227
00:14:17,280 --> 00:14:22,960
Speaker 1: something digitally because at the heart of digital information is

228
00:14:23,400 --> 00:14:28,120
Speaker 1: the bit, the basic unit of information. It is a

229
00:14:28,240 --> 00:14:31,840
Speaker 1: unit of information that only has two states zero or

230
00:14:32,000 --> 00:14:36,120
Speaker 1: one is essentially off or on. When you get down

231
00:14:36,160 --> 00:14:41,040
Speaker 1: to defining information in just two states, then you start

232
00:14:41,080 --> 00:14:44,040
Speaker 1: to look at something that is continuous and you realize

233
00:14:44,560 --> 00:14:46,240
Speaker 1: this is going to be a challenge. How do I

234
00:14:46,320 --> 00:14:52,160
Speaker 1: describe a continuous experience in very discrete amounts of information.

235
00:14:53,160 --> 00:14:57,280
Speaker 1: And that's when we get to the methodology we've developed

236
00:14:57,920 --> 00:15:01,280
Speaker 1: to digitally encode sound. I'm going to get into that

237
00:15:01,320 --> 00:15:04,680
Speaker 1: in just a minute, but before I do that, let's

238
00:15:04,720 --> 00:15:16,240
Speaker 1: take a quick break to thank our sponsor. All right,

239
00:15:16,360 --> 00:15:20,400
Speaker 1: let's get back into it. So we've talked about the

240
00:15:20,480 --> 00:15:23,400
Speaker 1: nature of sound. Analog sound, by the way, tries to

241
00:15:23,440 --> 00:15:27,560
Speaker 1: replicate exactly what we would experience in nature. It tries

242
00:15:27,560 --> 00:15:32,640
Speaker 1: to create this continuous experience, so you get these smooth

243
00:15:32,720 --> 00:15:38,320
Speaker 1: waves of frequencies and amplitudes. And that's why some people

244
00:15:38,480 --> 00:15:43,920
Speaker 1: argue that that analog styles of of sound recordings are

245
00:15:44,000 --> 00:15:48,800
Speaker 1: superior to digital ones. I don't necessarily think they're right,

246
00:15:49,360 --> 00:15:52,800
Speaker 1: but they often feel that way. So something like a

247
00:15:52,920 --> 00:15:58,000
Speaker 1: vinyl album, which is an analog format of digital or sorry,

248
00:15:58,040 --> 00:16:02,280
Speaker 1: an analog format of music storage should say sound storage. Uh,

249
00:16:02,320 --> 00:16:04,960
Speaker 1: they think that that is superior to say a c D,

250
00:16:05,320 --> 00:16:10,320
Speaker 1: which is a digital storage format. Uh. And who's to say.

251
00:16:10,400 --> 00:16:14,440
Speaker 1: I mean, like, if your sense of hearing is incredibly

252
00:16:14,720 --> 00:16:18,040
Speaker 1: well tuned, you might be able to pick up on

253
00:16:18,120 --> 00:16:22,160
Speaker 1: some differences. Or if someone did a really terrible job

254
00:16:22,680 --> 00:16:28,000
Speaker 1: encoding music digitally, then that might reveal itself to you

255
00:16:28,040 --> 00:16:30,760
Speaker 1: as well. Uh. But this is one of those things

256
00:16:30,800 --> 00:16:32,960
Speaker 1: that I think a lot of people feel they can

257
00:16:32,960 --> 00:16:34,760
Speaker 1: tell the difference, but if they would do a double

258
00:16:34,800 --> 00:16:39,360
Speaker 1: blind test, they might be surprised at how difficult it is.

259
00:16:39,840 --> 00:16:43,200
Speaker 1: If things if everything's working the way it should, then

260
00:16:43,440 --> 00:16:48,000
Speaker 1: there shouldn't be a perceptible difference at any rate. Digital

261
00:16:48,040 --> 00:16:54,360
Speaker 1: audio has two really important factors. Sample rate and bit depth,

262
00:16:55,160 --> 00:16:57,640
Speaker 1: or to another extent, bit rate. We'll talk about bit

263
00:16:57,760 --> 00:17:02,280
Speaker 1: rate as well. So the sample rate refers to how

264
00:17:02,320 --> 00:17:05,919
Speaker 1: many times you reference an analog sound to create the

265
00:17:05,960 --> 00:17:09,760
Speaker 1: digital version. So sound, like I said, is uninterrupted in

266
00:17:09,800 --> 00:17:14,840
Speaker 1: the analog world, you've got that that nice wave form.

267
00:17:14,920 --> 00:17:18,040
Speaker 1: In the analog world, that's not how digital world works.

268
00:17:18,119 --> 00:17:21,320
Speaker 1: Digital world, we have to describe that sound in a

269
00:17:21,400 --> 00:17:27,600
Speaker 1: series of discrete snippets of sound. It's probably easiest to

270
00:17:27,640 --> 00:17:33,840
Speaker 1: describe this with an analogy to movies on film. If

271
00:17:33,880 --> 00:17:37,359
Speaker 1: you work with film, like you're creating a movie on film,

272
00:17:37,840 --> 00:17:41,040
Speaker 1: then you know that you're not looking at a real

273
00:17:41,240 --> 00:17:44,240
Speaker 1: moving picture when you see the film played out at

274
00:17:44,280 --> 00:17:47,560
Speaker 1: the cinema. Instead, what you're looking at is a series

275
00:17:47,640 --> 00:17:52,160
Speaker 1: of photographs. If you take a film strip and you

276
00:17:52,240 --> 00:17:56,280
Speaker 1: look at it under a light, you'll see it's one

277
00:17:56,359 --> 00:18:00,760
Speaker 1: after another photograph. It's just a series of pictures. It's

278
00:18:00,760 --> 00:18:02,920
Speaker 1: only when you play them back at the right speed

279
00:18:03,520 --> 00:18:05,800
Speaker 1: and you projected onto a screen that you get the

280
00:18:05,880 --> 00:18:10,520
Speaker 1: illusion of continuous motion. But it's not really continuous. It's

281
00:18:10,560 --> 00:18:13,800
Speaker 1: just this series of photographs played at twenty four frames

282
00:18:13,800 --> 00:18:18,840
Speaker 1: per second in the case of actual film. So that

283
00:18:19,040 --> 00:18:22,200
Speaker 1: ends up being very analogous to the way we encode

284
00:18:22,200 --> 00:18:26,040
Speaker 1: digital audio. You take the analog recording and you take

285
00:18:26,359 --> 00:18:31,840
Speaker 1: snapshots of sound. The more frequently you take those snapshots,

286
00:18:32,240 --> 00:18:34,480
Speaker 1: the higher your sample rates. So in other words, if

287
00:18:34,480 --> 00:18:37,639
Speaker 1: you did one a second, your sample rate would be awful.

288
00:18:38,400 --> 00:18:40,600
Speaker 1: You would have a sample rate of one. But the

289
00:18:40,680 --> 00:18:43,680
Speaker 1: higher the sample rate, the closer your digital representation will

290
00:18:43,680 --> 00:18:47,280
Speaker 1: be to the frequency in the analog sound format. Actually,

291
00:18:47,760 --> 00:18:50,000
Speaker 1: what's really important to remember is that your sample rate

292
00:18:50,040 --> 00:18:52,480
Speaker 1: has to be about twice actually does have to be

293
00:18:52,520 --> 00:18:56,920
Speaker 1: twice what the highest frequency sound is in your recording.

294
00:18:58,440 --> 00:19:01,480
Speaker 1: It has to be because as if it's not, it

295
00:19:01,640 --> 00:19:07,720
Speaker 1: cannot encode that sound accurately. It's kind of interesting and

296
00:19:07,840 --> 00:19:09,919
Speaker 1: you might wonder, how do we take these snapshots in

297
00:19:09,920 --> 00:19:12,920
Speaker 1: the first place. Well, if you're capturing audio, let's say

298
00:19:12,960 --> 00:19:16,360
Speaker 1: we're recording to digital, So we've got a microphone set

299
00:19:16,440 --> 00:19:20,960
Speaker 1: up and we're recording to a digital media storage. Like

300
00:19:21,040 --> 00:19:23,280
Speaker 1: let's just say we're recording straight to someone's hard drive.

301
00:19:23,440 --> 00:19:26,760
Speaker 1: So we're talking into a microphone recording to a hard drive.

302
00:19:27,720 --> 00:19:31,440
Speaker 1: So you're using an analog microphone. Let's say you would

303
00:19:31,440 --> 00:19:35,760
Speaker 1: need an analog to digital converter Now this particular component

304
00:19:36,040 --> 00:19:40,800
Speaker 1: can receive discrete voltages from another device like your microphone.

305
00:19:41,040 --> 00:19:47,920
Speaker 1: So your microphone is converting sound into uh differences in voltage.

306
00:19:48,000 --> 00:19:50,880
Speaker 1: That's essentially how it communicates, so that it can then

307
00:19:51,040 --> 00:19:54,080
Speaker 1: send that to some other element. In this case, it's

308
00:19:54,119 --> 00:19:57,720
Speaker 1: sending it to the the analog to digital converter so

309
00:19:57,760 --> 00:20:00,400
Speaker 1: that it can be stored digitally on your our drive.

310
00:20:01,480 --> 00:20:08,560
Speaker 1: So this analog digital converters references or samples the discrete

311
00:20:08,680 --> 00:20:12,240
Speaker 1: voltage many times every second in order to create a

312
00:20:12,280 --> 00:20:16,760
Speaker 1: digital representation of the analog sound. It converts the voltages

313
00:20:16,840 --> 00:20:21,399
Speaker 1: into numbers and a process called quantization, and we express

314
00:20:21,480 --> 00:20:24,480
Speaker 1: those numbers in bits, So these are zeros and ones.

315
00:20:25,080 --> 00:20:27,760
Speaker 1: When you want to play the digital audio, a digital

316
00:20:27,800 --> 00:20:31,840
Speaker 1: to analog converter does the same process in reverse. So

317
00:20:32,080 --> 00:20:35,800
Speaker 1: it takes this digital information, these zeros and ones and

318
00:20:35,880 --> 00:20:39,600
Speaker 1: converts it into a series of discrete voltages, which then

319
00:20:39,840 --> 00:20:43,520
Speaker 1: can be amplified and sent to a speaker and create sound.

320
00:20:44,760 --> 00:20:47,360
Speaker 1: So all of that's really important. But now let's let's

321
00:20:47,359 --> 00:20:49,960
Speaker 1: talk about some concrete examples, and the best way to

322
00:20:49,960 --> 00:20:53,240
Speaker 1: do this is to go with compact discs. Because we

323
00:20:53,320 --> 00:20:57,119
Speaker 1: have a standard sample rate for compact discs, and that

324
00:20:57,280 --> 00:21:00,560
Speaker 1: standard sample rate is forty four point one la hurts

325
00:21:00,680 --> 00:21:04,200
Speaker 1: to create CD equality audio. That means that the audio

326
00:21:04,280 --> 00:21:10,000
Speaker 1: is sampled forty four thousand, one hundred times every second

327
00:21:10,880 --> 00:21:12,840
Speaker 1: the way they hear you say, the range of human

328
00:21:12,880 --> 00:21:15,359
Speaker 1: hearing you said only goes to twenty hurts to twenty

329
00:21:15,440 --> 00:21:18,280
Speaker 1: killer hurts. If it only goes up to twenty killer hurts,

330
00:21:18,280 --> 00:21:21,080
Speaker 1: why are you sampling at forty four thousand, one hundred

331
00:21:21,160 --> 00:21:25,560
Speaker 1: times every second? If it's twenty thousand times a second

332
00:21:25,600 --> 00:21:28,919
Speaker 1: for the frequency, why go up to four thousand, one

333
00:21:29,000 --> 00:21:31,520
Speaker 1: hundred Is there some relationship between that and the c

334
00:21:31,640 --> 00:21:34,680
Speaker 1: D sample rate? And the answer is yes. So there

335
00:21:34,800 --> 00:21:40,000
Speaker 1: is a theorem called the Nyquist Shannon sampling theorem, and

336
00:21:40,080 --> 00:21:42,760
Speaker 1: that states that the sample rate must be twice the

337
00:21:42,840 --> 00:21:46,000
Speaker 1: maximum frequency of a recording in order to describe the

338
00:21:46,040 --> 00:21:50,240
Speaker 1: frequency properly. So the general thought is the maximum frequency

339
00:21:50,320 --> 00:21:52,919
Speaker 1: most humans can here's twenty killer hurts. And for that reason,

340
00:21:52,960 --> 00:21:55,760
Speaker 1: Phillips and Sony when they were working to create the

341
00:21:55,960 --> 00:21:59,879
Speaker 1: CD format to make it a standard, they decide on

342
00:22:00,040 --> 00:22:02,879
Speaker 1: forty four point one killer hurts as that standard sample

343
00:22:03,000 --> 00:22:05,399
Speaker 1: rate for c D audio. It was more than double

344
00:22:05,440 --> 00:22:08,040
Speaker 1: the top frequency generally considered to be in the upper

345
00:22:08,119 --> 00:22:11,200
Speaker 1: level of human hearing. But what happens if you were

346
00:22:11,200 --> 00:22:14,400
Speaker 1: to lower the sampling rate. What if you didn't sample

347
00:22:14,480 --> 00:22:19,600
Speaker 1: at What if you sampled at let's say sixteen killer hurts,

348
00:22:19,600 --> 00:22:23,120
Speaker 1: so sixteen thousand times a second you sample it well,

349
00:22:23,400 --> 00:22:25,560
Speaker 1: that means you would only be able to record and

350
00:22:25,600 --> 00:22:29,200
Speaker 1: replicate any sound with a frequency up to eight killer

351
00:22:29,280 --> 00:22:34,280
Speaker 1: hurts or less, so eight thousand hurts or less. But

352
00:22:34,440 --> 00:22:37,640
Speaker 1: if you had any sound that was greater than eight

353
00:22:37,640 --> 00:22:42,080
Speaker 1: thousand hurts or eight killer hurts, anything higher than that,

354
00:22:43,080 --> 00:22:46,400
Speaker 1: it would be folded down to fit below the eight

355
00:22:46,480 --> 00:22:50,200
Speaker 1: killer hurts limit. Perceptually, that means the sounds you would

356
00:22:50,240 --> 00:22:53,199
Speaker 1: hear in the playback could include frequencies that were not

357
00:22:53,359 --> 00:22:58,160
Speaker 1: present in the original performance of that sound. So let's

358
00:22:58,160 --> 00:23:02,600
Speaker 1: say that I'm using a sample rate of sixteen uh,

359
00:23:02,640 --> 00:23:06,359
Speaker 1: you know, killer hurts, and someone is playing a musical

360
00:23:06,440 --> 00:23:09,200
Speaker 1: instrument and they play a note that's at a nine

361
00:23:09,280 --> 00:23:14,760
Speaker 1: killer hurts frequency. Well, because I'm sampling at sixteen killer hurts,

362
00:23:15,400 --> 00:23:19,679
Speaker 1: my limit for frequencies is eight killer hurts. If you

363
00:23:19,720 --> 00:23:22,600
Speaker 1: play something at nine killer hurts, what happens is it

364
00:23:22,920 --> 00:23:27,280
Speaker 1: the recording seems to fold the sound back, and it

365
00:23:27,400 --> 00:23:31,879
Speaker 1: folds it back at the same limit that the sound

366
00:23:31,960 --> 00:23:37,119
Speaker 1: goes over, the sample rate rather the Nyquist limit, I

367
00:23:37,119 --> 00:23:39,639
Speaker 1: should say, not the sample rateself, but the Nyquist limit.

368
00:23:40,760 --> 00:23:45,760
Speaker 1: So nine killer hurts sound played, My limit is eight

369
00:23:45,840 --> 00:23:49,000
Speaker 1: killer hurts. Well, nine killer hurts is one killer hurts

370
00:23:49,000 --> 00:23:52,040
Speaker 1: more than eight, so it folds it back and the

371
00:23:52,080 --> 00:23:55,359
Speaker 1: sound you would hear on the recording would be seven

372
00:23:55,440 --> 00:23:59,040
Speaker 1: killer hurts. So the original sound is nine killer hurts.

373
00:23:59,119 --> 00:24:03,480
Speaker 1: The playbacks sound is seven killer hurts, and you would

374
00:24:03,560 --> 00:24:07,719
Speaker 1: hear something recorded that wasn't actually played. That's why you

375
00:24:07,760 --> 00:24:10,840
Speaker 1: have to have a really high sample rate so that

376
00:24:10,880 --> 00:24:14,720
Speaker 1: you don't have these instances where sound gets folded back

377
00:24:15,520 --> 00:24:20,399
Speaker 1: into the frequency range, because otherwise what you were hearing

378
00:24:20,560 --> 00:24:24,560
Speaker 1: is not an accurate representation of what was actually generated

379
00:24:24,800 --> 00:24:28,960
Speaker 1: what you were trying to record. This whole phenomenon, by

380
00:24:29,000 --> 00:24:32,320
Speaker 1: the way, is called fold over or sometimes alias sing.

381
00:24:33,720 --> 00:24:36,880
Speaker 1: So that's sample rate. But then we've got bit depth. Now,

382
00:24:36,920 --> 00:24:41,159
Speaker 1: this is all about measuring the volume or amplitude of

383
00:24:41,160 --> 00:24:44,440
Speaker 1: a sound. So you have a range. You just make

384
00:24:44,440 --> 00:24:48,280
Speaker 1: an arbitrary range to say, like we're gonna go quietest

385
00:24:48,320 --> 00:24:51,320
Speaker 1: to loudest, and you just define what that range is.

386
00:24:51,440 --> 00:24:54,160
Speaker 1: It could literally be any range. Let's say you say

387
00:24:54,240 --> 00:24:58,360
Speaker 1: zero to one. Zero is dead silence, no sound at all.

388
00:24:58,840 --> 00:25:02,560
Speaker 1: One hundred is as loud as the sound ever gets.

389
00:25:02,680 --> 00:25:06,480
Speaker 1: It's the peak volume of sound. That means you can

390
00:25:06,560 --> 00:25:11,399
Speaker 1: describe all the different volumes within that recording at a

391
00:25:11,520 --> 00:25:15,359
Speaker 1: number between zero and one hundred. But let's say you

392
00:25:15,440 --> 00:25:18,800
Speaker 1: take that same recording and instead of making the range

393
00:25:19,000 --> 00:25:22,840
Speaker 1: zero to one hundred, you say it's zero to two thousand.

394
00:25:23,240 --> 00:25:26,840
Speaker 1: You haven't made the volume louder. The volume is still

395
00:25:27,080 --> 00:25:29,679
Speaker 1: the exact same as it was when you called the

396
00:25:29,760 --> 00:25:32,800
Speaker 1: range zero to one hundred. But what you have done

397
00:25:33,240 --> 00:25:38,160
Speaker 1: is added more units. You've created more precise steps between

398
00:25:38,400 --> 00:25:43,520
Speaker 1: absolute silent and as loud as it gets. So you've

399
00:25:43,560 --> 00:25:45,359
Speaker 1: just increased the size of the range so that you

400
00:25:45,400 --> 00:25:48,959
Speaker 1: can be more precise in the differences in volume. And

401
00:25:48,960 --> 00:25:52,440
Speaker 1: this is really important. So let's say that you've got

402
00:25:52,480 --> 00:25:55,280
Speaker 1: a sound that you rank at seventy eight and another

403
00:25:55,320 --> 00:25:58,800
Speaker 1: sound that you rank at seventy nine, and that's gonna

404
00:25:58,800 --> 00:26:01,159
Speaker 1: be the same for both of these changes. Uh, just

405
00:26:01,240 --> 00:26:04,000
Speaker 1: two different examples. Actually, So you've got your zero to

406
00:26:04,040 --> 00:26:08,000
Speaker 1: one range and a seventy eight would be seventy eight

407
00:26:08,080 --> 00:26:12,399
Speaker 1: percent of the loudest sound in the entire recording, and

408
00:26:12,440 --> 00:26:15,320
Speaker 1: at seventy nine would be a seventy nine of the

409
00:26:15,400 --> 00:26:19,320
Speaker 1: loudest sound in the entire recording. That's an actually pretty

410
00:26:19,320 --> 00:26:21,920
Speaker 1: hefty jump. But let's say we instead went with that

411
00:26:22,119 --> 00:26:25,320
Speaker 1: zero to two thousand range and you still had seventy

412
00:26:25,320 --> 00:26:29,600
Speaker 1: eight and seventy nine. Well, seventy eight would represent three

413
00:26:29,640 --> 00:26:33,160
Speaker 1: point nine percent of the full volume and seventy nine

414
00:26:33,160 --> 00:26:37,199
Speaker 1: would represent represent three point nine five of a full volume.

415
00:26:37,400 --> 00:26:40,520
Speaker 1: In other words, you'd be able to mark much more

416
00:26:40,640 --> 00:26:44,600
Speaker 1: subtle differences in volume, and that means you can have

417
00:26:44,680 --> 00:26:49,440
Speaker 1: more nuance in your recording. And since we're talking about

418
00:26:49,560 --> 00:26:52,280
Speaker 1: a natural sound to start off with, so you're taking

419
00:26:52,320 --> 00:26:55,560
Speaker 1: a natural sound and you're trying to digitize it. Smooth

420
00:26:55,640 --> 00:26:59,879
Speaker 1: changes in amplitude are possible in natural sound. Using a

421
00:27:00,000 --> 00:27:03,159
Speaker 1: broader range to describe the volume is best if you

422
00:27:03,200 --> 00:27:08,040
Speaker 1: want to get an accurate representation or resolution of that sound.

423
00:27:08,680 --> 00:27:11,679
Speaker 1: Going back to that zero to one range changes in

424
00:27:11,720 --> 00:27:15,000
Speaker 1: volume would be more chunky. Two sounds that have slight

425
00:27:15,080 --> 00:27:18,960
Speaker 1: differences in amplitude would end up being defined as being

426
00:27:18,960 --> 00:27:22,840
Speaker 1: identical because you wouldn't have the precision. You know, you

427
00:27:22,840 --> 00:27:25,320
Speaker 1: couldn't say this one seventy eight and a half. It

428
00:27:25,320 --> 00:27:27,680
Speaker 1: would either be seventy eight or seventy nine. So you

429
00:27:27,720 --> 00:27:31,520
Speaker 1: could have two sounds that in greater precision you could

430
00:27:31,520 --> 00:27:35,080
Speaker 1: tell the difference between their volumes. But if you have

431
00:27:35,359 --> 00:27:39,720
Speaker 1: that lower, that more shallow bit depth, you wouldn't be

432
00:27:39,760 --> 00:27:41,440
Speaker 1: able to tell the difference of it. You would lose

433
00:27:41,520 --> 00:27:44,760
Speaker 1: that nuance, that subtlety. This is part of the reason

434
00:27:44,880 --> 00:27:48,960
Speaker 1: why people say, like a lot of the modern music

435
00:27:49,080 --> 00:27:53,280
Speaker 1: has uh lower ranges and changes in volume, like the

436
00:27:53,280 --> 00:27:56,719
Speaker 1: the loudest loud parts and the softest soft parts. That

437
00:27:56,880 --> 00:28:00,800
Speaker 1: range has decreased over time, which a lot of people

438
00:28:00,800 --> 00:28:04,800
Speaker 1: have argued has meant that music has gotten less complex

439
00:28:05,040 --> 00:28:09,280
Speaker 1: and therefore, in some minds, less interesting. That's on a

440
00:28:09,359 --> 00:28:13,760
Speaker 1: related uh kind of philosophy to what I'm talking about here.

441
00:28:15,400 --> 00:28:19,480
Speaker 1: So you want to have those smaller steps between each

442
00:28:19,560 --> 00:28:23,879
Speaker 1: unit so you can create greater resolution, more smoothness to

443
00:28:23,920 --> 00:28:28,840
Speaker 1: the recorded audio. And it's actually the bit rate and

444
00:28:28,920 --> 00:28:32,880
Speaker 1: CD audio that will help make the sound seem smooth.

445
00:28:33,600 --> 00:28:36,159
Speaker 1: So if you ever listened to eight bit music, you know,

446
00:28:36,320 --> 00:28:39,040
Speaker 1: like the kind from old video game consoles. That sound

447
00:28:39,080 --> 00:28:42,520
Speaker 1: is really harsh and sort of chunky. It has an appeal,

448
00:28:42,800 --> 00:28:46,320
Speaker 1: but it's not you know, it's not smooth at all.

449
00:28:47,120 --> 00:28:49,680
Speaker 1: It can create an amazing effect, but if you want

450
00:28:49,720 --> 00:28:54,160
Speaker 1: to represent true analog sound, it's not awesome. But if

451
00:28:54,200 --> 00:28:58,800
Speaker 1: you went up to sixteen bit, that's CD quality bit depth,

452
00:28:59,480 --> 00:29:04,080
Speaker 1: it's much better. Uh, professional recording studios will do twenty

453
00:29:04,120 --> 00:29:07,400
Speaker 1: four bit or thirty two bit because they're gonna do

454
00:29:07,440 --> 00:29:10,840
Speaker 1: a lot of post processing work on those audio files.

455
00:29:11,120 --> 00:29:13,040
Speaker 1: And when you do that post processing work, if you

456
00:29:13,080 --> 00:29:16,880
Speaker 1: do it at sixteen bit, the stuff you're doing, the

457
00:29:16,960 --> 00:29:19,880
Speaker 1: changes you make, can become noticeable, and most times you

458
00:29:19,920 --> 00:29:22,480
Speaker 1: don't want that. You don't want it to be you know,

459
00:29:22,720 --> 00:29:24,640
Speaker 1: you don't want it to stand out from the rest

460
00:29:24,680 --> 00:29:27,360
Speaker 1: of the audio file. But that's the only reason they

461
00:29:27,400 --> 00:29:29,400
Speaker 1: go up to twenty four bit or thirty two bit.

462
00:29:29,760 --> 00:29:33,360
Speaker 1: There'd be no point in playing it back at that rate,

463
00:29:33,520 --> 00:29:39,240
Speaker 1: that bit depth, because human hearing is not so adept

464
00:29:39,360 --> 00:29:42,320
Speaker 1: to tell the difference, at least not for most humans.

465
00:29:43,200 --> 00:29:46,280
Speaker 1: So if you played back a recording at sixteen bit

466
00:29:46,640 --> 00:29:48,400
Speaker 1: and another one at twenty four bit, and it's the

467
00:29:48,480 --> 00:29:51,640
Speaker 1: same piece, most people would not be able to tell

468
00:29:51,640 --> 00:29:55,120
Speaker 1: the difference because you've already reached a resolution that equals

469
00:29:55,240 --> 00:29:59,120
Speaker 1: the precision of human hearing. Keeping in mind again, human

470
00:29:59,160 --> 00:30:02,080
Speaker 1: hearing is subject. If not everyone is equal, there's some

471
00:30:02,120 --> 00:30:05,640
Speaker 1: people who have incredible hearing who may be able to

472
00:30:05,680 --> 00:30:08,880
Speaker 1: pick out that difference. I am not one of those people,

473
00:30:09,480 --> 00:30:11,240
Speaker 1: but I am a person who's going to tell you.

474
00:30:11,760 --> 00:30:14,240
Speaker 1: We'll get to the last section in just a bit,

475
00:30:14,720 --> 00:30:18,440
Speaker 1: but first let's take another quick break to thank our sponsor.

476
00:30:27,120 --> 00:30:29,800
Speaker 1: All Right, so bids depth. What we just talked about

477
00:30:30,120 --> 00:30:32,400
Speaker 1: that can be thought of is how well the sound

478
00:30:32,480 --> 00:30:36,960
Speaker 1: is described, and the sampling rate is how frequently or

479
00:30:37,000 --> 00:30:41,640
Speaker 1: how much the sound is described. And CD Audio quality

480
00:30:41,680 --> 00:30:45,280
Speaker 1: has sixteen bit audio. That means that they actually have

481
00:30:45,520 --> 00:30:50,280
Speaker 1: sixty five thousand, five hundred thirty six different levels of

482
00:30:50,360 --> 00:30:55,120
Speaker 1: volume that they can describe within an audio track. So

483
00:30:55,200 --> 00:30:59,520
Speaker 1: my example of zero to two thousand that is primitive

484
00:30:59,640 --> 00:31:01,800
Speaker 1: compared at the c D audio because it has the

485
00:31:01,960 --> 00:31:06,120
Speaker 1: sixteen bit style six five hundred thirty six different levels.

486
00:31:06,880 --> 00:31:11,240
Speaker 1: And how is that possible? Well, when we say sixteen bit,

487
00:31:11,920 --> 00:31:15,320
Speaker 1: remember a bit represents two states zero or one. So

488
00:31:15,360 --> 00:31:18,720
Speaker 1: you take the number two and then you raise it

489
00:31:18,760 --> 00:31:23,680
Speaker 1: to the power of sixteen. Uh, so you multiply to

490
00:31:24,000 --> 00:31:27,200
Speaker 1: by itself sixteen times and you get sixty five thousand,

491
00:31:27,200 --> 00:31:30,240
Speaker 1: three D fifty six. So that's that's where that number

492
00:31:30,280 --> 00:31:34,440
Speaker 1: comes from. Now, with your digital sample, you have a

493
00:31:34,480 --> 00:31:37,640
Speaker 1: collection of points that roughly replicate the shape of an

494
00:31:37,680 --> 00:31:40,680
Speaker 1: analog sound wave. It's gonna look a little funky, but

495
00:31:41,120 --> 00:31:44,720
Speaker 1: you'll be able to see what the frequency and amplitude

496
00:31:45,120 --> 00:31:48,240
Speaker 1: generally was of the original recording if you were to

497
00:31:48,400 --> 00:31:51,960
Speaker 1: plot this on an X y axis. But if you

498
00:31:52,000 --> 00:31:55,480
Speaker 1: were just to connect each successive point with a straight line,

499
00:31:56,080 --> 00:31:58,680
Speaker 1: even as close together as they would be, because you're

500
00:31:58,720 --> 00:32:02,160
Speaker 1: looking at forty four thousand one times a second, it

501
00:32:02,200 --> 00:32:05,560
Speaker 1: had sound pretty awful. So we actually use an algorithm

502
00:32:05,640 --> 00:32:10,240
Speaker 1: called interpolation to join the points smoothly to imitate a

503
00:32:10,320 --> 00:32:13,600
Speaker 1: sound wave form, and that gives a musical playback program

504
00:32:13,640 --> 00:32:17,280
Speaker 1: the ability to replicate an analog wave form. And that's

505
00:32:17,320 --> 00:32:22,240
Speaker 1: actually called pulse code modulation or pc M. And if

506
00:32:22,280 --> 00:32:27,760
Speaker 1: you store audio uh intact this way, you would have

507
00:32:27,840 --> 00:32:31,920
Speaker 1: what we call a lossless audio file, which means exactly

508
00:32:31,920 --> 00:32:34,880
Speaker 1: what it sounds like. None of that data would ever

509
00:32:34,920 --> 00:32:37,800
Speaker 1: get filtered out of the file, even if the sounds

510
00:32:37,880 --> 00:32:40,320
Speaker 1: were beyond the range of human hearing, they would be

511
00:32:40,360 --> 00:32:45,120
Speaker 1: recorded and you would have a lossless file format. Those

512
00:32:45,160 --> 00:32:47,840
Speaker 1: files tend to be quite big, depending upon how long

513
00:32:47,840 --> 00:32:51,600
Speaker 1: a recording you make, of course. All right, so now

514
00:32:52,040 --> 00:32:53,720
Speaker 1: here's where it gets a little confusing. And I think

515
00:32:53,720 --> 00:32:55,520
Speaker 1: I even said bit rate a couple of times when

516
00:32:55,520 --> 00:32:58,720
Speaker 1: I really meant bit depths earlier. But up to this point,

517
00:32:58,760 --> 00:33:02,680
Speaker 1: I really was talking at depth. So my apologies to

518
00:33:02,720 --> 00:33:05,280
Speaker 1: all of you out there if a bit rate slipped through,

519
00:33:05,760 --> 00:33:07,360
Speaker 1: because I did not mean it. Now I'm going to

520
00:33:07,440 --> 00:33:10,480
Speaker 1: talk about bit rate and show you how it's different

521
00:33:10,520 --> 00:33:14,920
Speaker 1: than bit depth. Bit Rate refers to the amount of

522
00:33:15,040 --> 00:33:20,000
Speaker 1: data audio uses per second or requires per second of recording,

523
00:33:20,560 --> 00:33:23,920
Speaker 1: and you derive bit rate from the bit depth and

524
00:33:24,000 --> 00:33:28,880
Speaker 1: the sampling rate. It's represented as bits per second. So again,

525
00:33:28,920 --> 00:33:31,120
Speaker 1: let's go to ceed equality sound. That makes it easy.

526
00:33:31,400 --> 00:33:36,560
Speaker 1: You have thousand one samples per second. You've got sixteen

527
00:33:36,640 --> 00:33:40,959
Speaker 1: bits or two bites, because remember a bite is eight bits,

528
00:33:41,760 --> 00:33:45,480
Speaker 1: so you've got two bites to describe each sample. So

529
00:33:45,680 --> 00:33:51,239
Speaker 1: two bites for one samples per second. Uh plus you

530
00:33:51,400 --> 00:33:54,360
Speaker 1: probably are gonna have to multiply that by two because

531
00:33:54,400 --> 00:33:57,120
Speaker 1: you're probably recording in stereo, so you have to do

532
00:33:57,160 --> 00:34:01,560
Speaker 1: that once reach track, so you get that number, then

533
00:34:01,600 --> 00:34:04,120
Speaker 1: you have to multiply that by sixty seconds to determine

534
00:34:04,160 --> 00:34:07,360
Speaker 1: how much data per minute you are creating when you're recording,

535
00:34:07,840 --> 00:34:10,440
Speaker 1: and with seed quality audio, that ends up being about

536
00:34:10,480 --> 00:34:14,480
Speaker 1: ten megabytes of data per minute. Now these days that's

537
00:34:14,640 --> 00:34:17,960
Speaker 1: not really that big a deal because we're dealing with

538
00:34:18,080 --> 00:34:22,640
Speaker 1: super fast internet speeds and enormous hard drives. But just

539
00:34:22,880 --> 00:34:25,640
Speaker 1: a few years ago, that was considered to be a

540
00:34:25,840 --> 00:34:29,880
Speaker 1: really sizeable file, I mean an enormous file, and so

541
00:34:30,040 --> 00:34:32,480
Speaker 1: if you wanted to find a way to distribute digital

542
00:34:32,520 --> 00:34:35,560
Speaker 1: audio so it didn't take up too much space, you

543
00:34:35,719 --> 00:34:39,680
Speaker 1: had to figure out how you could compress those files

544
00:34:40,239 --> 00:34:43,799
Speaker 1: and make them smaller, make them more manageable. And now

545
00:34:43,840 --> 00:34:48,319
Speaker 1: we can finally get back to Germany and Hair Brandenburg.

546
00:34:49,000 --> 00:34:52,120
Speaker 1: You thought we left him behind, We didn't. He was

547
00:34:52,200 --> 00:34:55,480
Speaker 1: just part of a flashback. So let's go to the

548
00:34:55,560 --> 00:34:58,719
Speaker 1: MP three. First of all, it gets his name from

549
00:34:58,760 --> 00:35:02,120
Speaker 1: the Motion Picture at Spurts Group, also known as IMPEG.

550
00:35:03,160 --> 00:35:06,640
Speaker 1: It was part of a project that IMPEG was doing

551
00:35:06,680 --> 00:35:10,080
Speaker 1: that was looking at ways of compressing audio. Along with

552
00:35:10,840 --> 00:35:14,160
Speaker 1: the work that they were doing with video files. It's

553
00:35:14,160 --> 00:35:18,600
Speaker 1: actually named after the process that they developed, called IMPEG

554
00:35:18,640 --> 00:35:21,759
Speaker 1: Audio Layer three. So yes, there was a layer one

555
00:35:21,800 --> 00:35:25,120
Speaker 1: and a layer two. Layer three was a refinement of

556
00:35:25,160 --> 00:35:27,880
Speaker 1: the approach and was the one that was actually successful

557
00:35:27,920 --> 00:35:32,560
Speaker 1: in the market. Now, Brandenburg was working with an instructor

558
00:35:32,800 --> 00:35:36,040
Speaker 1: he was pursuing Brandenburg was pursuing a PhD at the

559
00:35:36,080 --> 00:35:38,680
Speaker 1: time and trying to come up with a practical means

560
00:35:38,680 --> 00:35:42,160
Speaker 1: of transmitting digital audio across phone lines, and in the

561
00:35:42,200 --> 00:35:45,600
Speaker 1: process he began to experiment with algorithms that could take

562
00:35:45,760 --> 00:35:51,280
Speaker 1: digital audio information and determine which bits are significant. Anything

563
00:35:51,320 --> 00:35:55,480
Speaker 1: that was deemed insignificant could be discarded. So the thinking

564
00:35:55,600 --> 00:35:59,440
Speaker 1: was that information we cannot perceive as human beings is worthless.

565
00:35:59,480 --> 00:36:03,320
Speaker 1: There's no point in preserving it in an audio file format.

566
00:36:03,360 --> 00:36:06,120
Speaker 1: It's just taking up space that we can't even perceive

567
00:36:06,200 --> 00:36:08,840
Speaker 1: when we play it back, So there's no reason to

568
00:36:08,880 --> 00:36:11,719
Speaker 1: replicate it, there's no reason to record it. Leave it out,

569
00:36:12,200 --> 00:36:15,560
Speaker 1: and that way you could compress digital audio files. Or

570
00:36:15,640 --> 00:36:18,800
Speaker 1: to put it another way, if the algorithm determined that

571
00:36:18,880 --> 00:36:21,440
Speaker 1: a sound was outside the range of human hearing, it

572
00:36:21,480 --> 00:36:24,399
Speaker 1: would drop it from the encoding process, so you get

573
00:36:24,440 --> 00:36:29,120
Speaker 1: a sound file much smaller than the more accurate representative version.

574
00:36:29,400 --> 00:36:32,520
Speaker 1: So the lossless version would be more accurate to the

575
00:36:32,560 --> 00:36:36,239
Speaker 1: original sound. But this new version, what we would call

576
00:36:36,280 --> 00:36:39,759
Speaker 1: a lossy version, a compressed file, would be able to

577
00:36:39,800 --> 00:36:44,480
Speaker 1: replicate it pretty well if it's designed properly, and maybe

578
00:36:44,880 --> 00:36:47,440
Speaker 1: to a point if you design it well enough that

579
00:36:47,520 --> 00:36:50,200
Speaker 1: you couldn't tell the difference between the two. Uh. That

580
00:36:50,280 --> 00:36:54,000
Speaker 1: took some time. That was not easy to do. So

581
00:36:55,360 --> 00:36:58,440
Speaker 1: the new file, the new version, the compressed one, the

582
00:36:58,480 --> 00:37:02,640
Speaker 1: lossy format, would only have the actual relevant data, and

583
00:37:02,719 --> 00:37:05,799
Speaker 1: from that point forward, the challenge was to determine what

584
00:37:05,920 --> 00:37:10,400
Speaker 1: are the benchmarks to figure out what is relevant versus

585
00:37:10,440 --> 00:37:13,719
Speaker 1: what is irrelevant, Because if you lose too much information,

586
00:37:13,800 --> 00:37:16,640
Speaker 1: you change the quality of the recording, meaning it's no

587
00:37:16,719 --> 00:37:20,440
Speaker 1: longer an accurate representation of the original sound. So you

588
00:37:20,520 --> 00:37:24,480
Speaker 1: might say that any sound below twenty hurts isn't relevant

589
00:37:24,520 --> 00:37:28,240
Speaker 1: because it's below the range of your typical human humans

590
00:37:28,280 --> 00:37:31,800
Speaker 1: ability to hear. You might say that anything above twenty

591
00:37:31,840 --> 00:37:37,280
Speaker 1: thousand hurts or twenty killer hurts is irrelevant because humans

592
00:37:37,280 --> 00:37:41,600
Speaker 1: typically can't hear sounds above that frequency. You might say

593
00:37:41,640 --> 00:37:46,120
Speaker 1: that sounds at a certain amplitude or lower are irrelevant

594
00:37:46,160 --> 00:37:50,360
Speaker 1: because they're so quiet that humans wouldn't hear them. Or

595
00:37:50,480 --> 00:37:53,560
Speaker 1: you might say that if a certain sound is at

596
00:37:53,600 --> 00:37:56,239
Speaker 1: a lower amplitude and a different sound is at a

597
00:37:56,320 --> 00:38:00,440
Speaker 1: higher amplitude, the higher amplitude sound is drowning out the

598
00:38:00,520 --> 00:38:04,120
Speaker 1: lower amplitude sound, and so we humans don't really perceive

599
00:38:04,200 --> 00:38:08,319
Speaker 1: the lower amplitude sound. This is where we get into psychoacoustics.

600
00:38:08,400 --> 00:38:10,880
Speaker 1: It's not just what we hear, but how we perceive

601
00:38:11,320 --> 00:38:15,120
Speaker 1: the sound itself. And a lot of that went into

602
00:38:15,320 --> 00:38:18,160
Speaker 1: formulating the algorithms to figure out how to compress this

603
00:38:18,280 --> 00:38:21,680
Speaker 1: music in a way where you get a recording that

604
00:38:22,120 --> 00:38:27,200
Speaker 1: represents the original without you know, compromising too much and

605
00:38:27,280 --> 00:38:31,400
Speaker 1: still getting the file size to a manageable size. And

606
00:38:31,440 --> 00:38:33,560
Speaker 1: these are the decisions you have to make to figure

607
00:38:33,560 --> 00:38:35,879
Speaker 1: out which bits of information you keep in which ones

608
00:38:35,920 --> 00:38:40,160
Speaker 1: you ditch. Brandenburg and a team we're working on refining

609
00:38:40,160 --> 00:38:44,120
Speaker 1: this approach in the late eighties and early nineties, And

610
00:38:44,120 --> 00:38:46,040
Speaker 1: he said, at one point he thought he had nailed it,

611
00:38:46,360 --> 00:38:49,720
Speaker 1: and then he heard an acapella song. It was Tom's

612
00:38:49,760 --> 00:38:54,400
Speaker 1: Diner by Suzanne Vega, And then he listened to the

613
00:38:54,400 --> 00:38:58,400
Speaker 1: compressed MP three version of that song using the the

614
00:38:58,600 --> 00:39:00,920
Speaker 1: version of MP three that had been developed up to

615
00:39:01,080 --> 00:39:04,800
Speaker 1: that point, and he said, it ruined the song. It

616
00:39:05,000 --> 00:39:09,800
Speaker 1: trashed it. It sounded terrible. He said that other representations

617
00:39:09,800 --> 00:39:13,160
Speaker 1: of music seemed fine with this particular approach, but when

618
00:39:13,200 --> 00:39:16,239
Speaker 1: they went with this stripped down acapella song with this

619
00:39:16,400 --> 00:39:19,840
Speaker 1: particular kind of you're in the middle of a space

620
00:39:19,920 --> 00:39:24,480
Speaker 1: listening to Suzanne Vegas sing, it ruined her voice. And

621
00:39:24,560 --> 00:39:27,080
Speaker 1: so the team began to tweet the compression algorithms to

622
00:39:27,160 --> 00:39:30,080
Speaker 1: correct for this problem, and it took a lot of

623
00:39:30,120 --> 00:39:33,520
Speaker 1: work to figure out, Okay, well, what are the elements

624
00:39:33,560 --> 00:39:37,440
Speaker 1: of sound that we messed with that have created this issue,

625
00:39:37,600 --> 00:39:39,839
Speaker 1: and ultimately they were finally able to create an MP

626
00:39:39,920 --> 00:39:43,400
Speaker 1: three file that didn't distort or ruin the recording. Brandberg

627
00:39:43,480 --> 00:39:46,160
Speaker 1: said he listened to that song somewhere between five hundred

628
00:39:46,200 --> 00:39:49,400
Speaker 1: and a thousand times, and then he saw Suzanne Vega

629
00:39:49,480 --> 00:39:53,640
Speaker 1: performance live and he was able to recognize all of

630
00:39:53,680 --> 00:39:58,760
Speaker 1: those subtle changes in her voice because he had paid

631
00:39:58,920 --> 00:40:01,720
Speaker 1: so close attention to it during the process of tweaking

632
00:40:01,719 --> 00:40:05,880
Speaker 1: this algorithm. He said, ultimately, the real telling thing is

633
00:40:06,040 --> 00:40:10,960
Speaker 1: he still enjoyed the song, which says a lot about him. Me.

634
00:40:11,120 --> 00:40:14,520
Speaker 1: I can't stand that song, but maybe it's just because

635
00:40:14,520 --> 00:40:16,400
Speaker 1: to me there's a point where it just sounds like

636
00:40:16,400 --> 00:40:18,960
Speaker 1: someone is just singing about what they're doing. And I

637
00:40:19,040 --> 00:40:22,960
Speaker 1: do that every day. No one gave me a record deal, alright.

638
00:40:22,960 --> 00:40:27,360
Speaker 1: So getting back to MP three, they had finalized the

639
00:40:27,760 --> 00:40:31,359
Speaker 1: foul format and created the standard, but it was just

640
00:40:31,520 --> 00:40:35,240
Speaker 1: one of several possibilities for encoding audio and it didn't

641
00:40:35,280 --> 00:40:42,040
Speaker 1: immediately take off. It wasn't immediately adopted by consumers. The

642
00:40:42,080 --> 00:40:45,799
Speaker 1: team had identified the Internet as a possible distribute distribution

643
00:40:45,840 --> 00:40:48,920
Speaker 1: method for MP three files, rather than just over telephone lines.

644
00:40:48,960 --> 00:40:51,520
Speaker 1: They said, well, can technically we could send and B

645
00:40:51,680 --> 00:40:56,240
Speaker 1: three's across the Internet, so you could send manageable sized

646
00:40:56,320 --> 00:41:02,120
Speaker 1: files across this network. Until life fourteen, they created the

647
00:41:02,160 --> 00:41:06,759
Speaker 1: file extension DOT MP three. Now it would take a

648
00:41:06,840 --> 00:41:09,960
Speaker 1: little bit longer for software to take advantage of this.

649
00:41:10,080 --> 00:41:13,359
Speaker 1: One of the early programs was win amp, which made

650
00:41:13,440 --> 00:41:16,479
Speaker 1: MP three decoding accessible, and from that point the file

651
00:41:16,560 --> 00:41:20,560
Speaker 1: format began to take off. To follow would be dedicated

652
00:41:20,640 --> 00:41:23,760
Speaker 1: MP three players and sites that allowed people to upload

653
00:41:23,800 --> 00:41:28,760
Speaker 1: and download compressed audio files, which also indicated a rise

654
00:41:28,880 --> 00:41:33,440
Speaker 1: in piracy, and then in response to the rise in piracy.

655
00:41:33,520 --> 00:41:36,400
Speaker 1: We saw an increase in d r M strategies digital

656
00:41:36,480 --> 00:41:40,960
Speaker 1: rights management or copy protection if you prefer, and that

657
00:41:41,120 --> 00:41:44,120
Speaker 1: all really ended up shaping a lot of the policies

658
00:41:44,320 --> 00:41:48,680
Speaker 1: and strategies that affect the Internet today. So you could

659
00:41:48,680 --> 00:41:51,720
Speaker 1: say that the MP three is one of the reasons

660
00:41:51,760 --> 00:41:54,440
Speaker 1: why the Internet is the way it is right now,

661
00:41:54,480 --> 00:41:58,799
Speaker 1: and why arguments both for and against net neutrality have

662
00:41:59,560 --> 00:42:02,080
Speaker 1: formula aided in certain ways. A lot of it is

663
00:42:02,120 --> 00:42:06,080
Speaker 1: shaped by the MP three. So that kind of wraps

664
00:42:06,160 --> 00:42:10,000
Speaker 1: up this discussion about digital audio in general and a

665
00:42:10,040 --> 00:42:12,640
Speaker 1: little bit on MP three files. In the next episode

666
00:42:12,640 --> 00:42:15,880
Speaker 1: of this series, I will dive into a more technical

667
00:42:15,960 --> 00:42:18,919
Speaker 1: explanation of what is actually going on with the MP

668
00:42:19,040 --> 00:42:23,239
Speaker 1: three compression algorithms, and I bet you can't wait to

669
00:42:23,320 --> 00:42:27,440
Speaker 1: learn all about fast Furrier transforms. I know I can't,

670
00:42:28,040 --> 00:42:31,040
Speaker 1: And like I said, I have other episodes to sprinkle

671
00:42:31,080 --> 00:42:33,440
Speaker 1: in between this one and the next one and then

672
00:42:33,520 --> 00:42:36,000
Speaker 1: the third one, so that way you won't just get

673
00:42:36,080 --> 00:42:39,560
Speaker 1: digital audio overload. And if you guys have any comments

674
00:42:39,760 --> 00:42:43,000
Speaker 1: or questions or suggestions for show topics or people I

675
00:42:43,040 --> 00:42:45,879
Speaker 1: should interview, or maybe people I should have on as

676
00:42:45,880 --> 00:42:49,560
Speaker 1: a guest host shoot him my way. My email is

677
00:42:49,640 --> 00:42:53,040
Speaker 1: tech stuff at how stuff works dot com, or you

678
00:42:53,040 --> 00:42:55,600
Speaker 1: can always drop me a line on Facebook or Twitter

679
00:42:55,760 --> 00:42:59,080
Speaker 1: with the handle tech stuff hs W and I'll talk

680
00:42:59,120 --> 00:43:05,719
Speaker 1: to you guys again really soon. For more on this

681
00:43:05,880 --> 00:43:08,399
Speaker 1: and thousands of other topics, is it how stuff works

682
00:43:08,400 --> 00:43:18,600
Speaker 1: dot com