WEBVTT Kind: captions Language: en 00:00:00.060 --> 00:00:03.740 Have you ever noticed that a loudspeaker is the opposite of an eardrum? 00:00:04.180 --> 00:00:05.320 Eh, probably not. 00:00:05.320 --> 00:00:06.390 But it’s true! 00:00:06.390 --> 00:00:10.809 See our ears work by concentrating changes in air pressure onto a small diaphragm that 00:00:10.809 --> 00:00:13.710 will move back and forth with the pressure changes. 00:00:13.710 --> 00:00:18.560 This vibration causes stimulation in the heary bits of the ear which your brain can, assuming 00:00:18.560 --> 00:00:22.730 you have normal hearing ability, turn into what we perceive as sound. 00:00:22.730 --> 00:00:27.720 A loudspeaker does the opposite--its diaphragms (the driver cones) vibrate to create pressure 00:00:27.720 --> 00:00:29.160 changes in the air. 00:00:29.160 --> 00:00:32.629 This vibration gets transferred to our eardrums so we can hear it. 00:00:32.629 --> 00:00:36.230 We’re sticking to simple stuff today because the rabbit hole is just too deep. 00:00:36.230 --> 00:00:39.860 All you need to know is that things vibrate, which causes air pressure to fluctuate, which 00:00:39.860 --> 00:00:44.040 causes our eardrums to also vibrate, which stimulates the brain so that we can perceive 00:00:44.040 --> 00:00:46.579 that vibration as sound. 00:00:46.579 --> 00:00:50.670 This channel started as a series exploring the history of artificial sound, and it’s 00:00:50.670 --> 00:00:54.719 been over TWO YEARS since I last touched on it at all. 00:00:54.719 --> 00:00:57.640 Finally we’re finishing this up with the introduction of 00:00:57.640 --> 00:01:00.000 DIGITAL SOUND (emphasis added with obnoxious reverb). 00:01:00.000 --> 00:01:04.860 Since it’s been forever, let’s go over a brief history of sound recording technologies. 00:01:04.869 --> 00:01:08.500 The first device which could reproduce a sound recording was the phonograph. 00:01:08.500 --> 00:01:12.650 Thomas Edison’s invention consisted of an artificial eardrum, which would vibrate along 00:01:12.650 --> 00:01:17.290 with changes in sound pressure, and with the aid of a collecting horn, the vibration is 00:01:17.290 --> 00:01:21.190 transferred into this stylus, creating an up-and-down motion. 00:01:21.190 --> 00:01:25.479 This carves a groove into a wax cylinder, and the vibrating stylus creates an imprint 00:01:25.479 --> 00:01:26.910 of the sound wave. 00:01:26.910 --> 00:01:32.049 The depth of that groove becomes a literal analog of the original sound vibrations. 00:01:32.049 --> 00:01:36.880 Then, when the stylus is run over the now bumpy groove, the bumps cause the diaphragm 00:01:36.880 --> 00:01:41.460 to vibrate in the same way as it did when it first made the bumps, and the result is 00:01:41.460 --> 00:01:44.170 that you hear the same sound as before. 00:01:44.170 --> 00:01:48.660 Or at least, a barely passable imitation of that sound. 00:01:49.620 --> 00:01:54.740 (sad sounding violin music) 00:02:02.740 --> 00:02:06.119 Commercially produced discs and cylinders were molded from master 00:02:06.119 --> 00:02:09.570 recordings, and wouldn’t wear down like the original wax cylinders. 00:02:09.570 --> 00:02:12.070 They were played back using devices like this. 00:02:12.070 --> 00:02:15.830 This device is called a reproducer, and for decades all phonographs were based on 00:02:15.830 --> 00:02:18.380 simple acoustic devices like this. 00:02:18.380 --> 00:02:22.860 For nearly a century, this is how artificial sound recording technologies worked. 00:02:22.860 --> 00:02:28.040 Something (like this horn) would collect sound waves, and recreate them onto a physical analog. 00:02:28.050 --> 00:02:33.700 Then, that physical analog could recreate the original sound waves when played back. 00:02:33.700 --> 00:02:38.440 While it all started with simple acoustic devices like this phonograph, eventually improvements 00:02:38.440 --> 00:02:39.440 were made. 00:02:39.440 --> 00:02:43.290 The development of the electronic microphone was perhaps the most important. 00:02:43.290 --> 00:02:48.810 Now, sound waves cause a receiving diaphragm to move a coil of wire around a magnet, and 00:02:48.810 --> 00:02:52.510 a voltage is produced in the wire as the diaphragm moves. 00:02:52.510 --> 00:02:56.660 This time, sound waves are recreated as a voltage coming from the microphone, and by 00:02:56.660 --> 00:03:01.250 amplifying this voltage and sending it into a new record cutting device which moves its 00:03:01.250 --> 00:03:05.830 cutting stylus as a function of the voltage it receives, a more accurate carving of the 00:03:05.830 --> 00:03:09.050 sound wave could be made into a disc or cylinder. 00:03:09.050 --> 00:03:13.290 This greatly improved the fidelity of the recorded sound, even on acoustic reproduction 00:03:13.290 --> 00:03:14.840 devices like this. 00:03:14.840 --> 00:03:19.340 With the proliferation of radio--which I feel I must explain is a sound transmission technology, 00:03:19.340 --> 00:03:20.740 not sound recording. 00:03:20.740 --> 00:03:23.160 Just so we don’t get confused too much here-- 00:03:23.160 --> 00:03:25.860 the loudspeaker became a big deal. 00:03:25.860 --> 00:03:30.960 Loudspeakers are the opposite of microphones--instead of producing a voltage as a reaction to a 00:03:30.970 --> 00:03:35.590 sound pressure wave moving its diaphragm, a loudspeaker will move its diaphragm and 00:03:35.590 --> 00:03:39.260 create a pressure wave as a reaction to incoming voltage. 00:03:39.260 --> 00:03:44.030 With loudspeakers all the rage, record players could now use a phonograph cartridge, which 00:03:44.030 --> 00:03:46.460 acts like a microphone for records. 00:03:46.460 --> 00:03:51.650 The movement of the stylus as the groove vibrates it generates a voltage which can be amplified 00:03:51.650 --> 00:03:53.380 to drive a loudspeaker. 00:03:53.380 --> 00:03:56.080 This gets very meta very quickly. 00:03:56.090 --> 00:04:00.520 An artificial ear turns sounds into voltage, and a cutting stylus turns this voltage into 00:04:00.520 --> 00:04:01.990 a groove on a record. 00:04:01.990 --> 00:04:07.840 Then, a playback stylus playing the record generates a voltage as the stylus vibrates. 00:04:07.840 --> 00:04:12.040 This voltage is then amplified to drive a loudspeaker, which causes pressure changes 00:04:12.040 --> 00:04:16.390 in the air around the loudspeaker, which your ears concentrate down to your eardrums, and 00:04:16.390 --> 00:04:21.070 now your real eardrums are vibrating in roughly the same way that the original artificial 00:04:21.070 --> 00:04:24.330 eardrum moved in the microphone in the first place. 00:04:24.330 --> 00:04:25.330 Yeah. 00:04:25.330 --> 00:04:30.350 In essence, the record becomes a way to recreate the original pattern of voltage created by 00:04:30.350 --> 00:04:34.480 the microphone, so that the sound can be heard again in a different place 00:04:34.480 --> 00:04:36.260 at a different time. 00:04:36.260 --> 00:04:38.880 Let’s cut out the middle bit because that’s what’s most confusing. 00:04:38.880 --> 00:04:42.930 A microphone like this creates an electrical signal of fluctuating intensity based on how 00:04:42.930 --> 00:04:44.690 its diaphragm moves. 00:04:44.690 --> 00:04:48.930 I can just amplify that signal and send it straight into a loudspeaker, which will reproduce 00:04:48.930 --> 00:04:50.840 the sound in real time. 00:04:50.840 --> 00:04:54.400 Radio accomplishes this wirelessly, but the sound isn’t recorded. 00:04:54.400 --> 00:04:59.300 To capture the sound coming from the microphone to be played back later, it has to be converted 00:04:59.300 --> 00:05:01.770 into an analog of the signal. 00:05:01.770 --> 00:05:04.900 And that’s why it’s called analog recording technology. 00:05:04.900 --> 00:05:09.820 No matter if it’s a record, a cassette tape, an open reel tape, or even a wax cylinder, 00:05:09.820 --> 00:05:12.370 the sound information is recorded “doorectly”... 00:05:12.370 --> 00:05:13.370 Doorectly. 00:05:13.370 --> 00:05:14.370 Doorectly? 00:05:14.370 --> 00:05:18.470 The sound information is recorded directly onto something, which can then be used to 00:05:18.470 --> 00:05:22.110 recreate a copy of the original sound information. 00:05:22.110 --> 00:05:25.720 That something is an analog of the original sound waves. 00:05:25.720 --> 00:05:30.100 Improvements in sound technology were for many years simply incremental. 00:05:30.110 --> 00:05:32.530 Wax cylinders became shellac discs. 00:05:32.530 --> 00:05:34.110 Shellac became vinyl. 00:05:34.110 --> 00:05:38.100 Magnetic recording wire allowed for a reusable, electronic recording medium. 00:05:38.100 --> 00:05:42.600 This was improved into magnetic tape, allowing for a high fidelity, versatile recording medium 00:05:42.600 --> 00:05:45.340 enabling multi-track recording and editing. 00:05:45.340 --> 00:05:49.400 And to improve on the noise of magnetic tape, different particle formulations were developed, 00:05:49.400 --> 00:05:51.540 and noise reduction technologies matured. 00:05:51.540 --> 00:05:55.810 But we were still just taking some signal from a microphone, then slapping it basically 00:05:55.810 --> 00:05:59.040 as is onto some sort of physical medium. 00:05:59.040 --> 00:06:01.420 And that medium was never perfect. 00:06:01.420 --> 00:06:04.490 Poorly made tape would cause signal dropouts. 00:06:04.490 --> 00:06:09.500 Discs would be plagued by dust and scratches, and would slowly wear down with each play. 00:06:09.500 --> 00:06:14.010 Because the analog medium contained the sound in its physical properties, it was inherently 00:06:14.010 --> 00:06:16.770 prone to wear, damage, and distortion. 00:06:16.770 --> 00:06:21.680 Which of course would wear down, damage, or distort the sound recording itself. 00:06:21.680 --> 00:06:26.670 If only there were some way to encode the sound, perhaps a way to store sound logically 00:06:26.670 --> 00:06:28.680 rather than analogously. 00:06:28.680 --> 00:06:32.910 Maybe if the signal weren’t the sound itself, but instead were a set of instructions on 00:06:32.910 --> 00:06:37.600 how to recreate it, we could get lossless, near-perfect sound reproduction. 00:06:37.600 --> 00:06:40.010 And thus, digital sound was born. 00:06:40.010 --> 00:06:44.770 The heart of uncompressed digital sound is pulse-code modulation, or PCM. 00:06:44.770 --> 00:06:49.840 PCM’s roots can be traced back to the telegraph days, but its invention as we know it today 00:06:49.840 --> 00:06:53.510 for sound came from British Engineer Alec Reeves. 00:06:53.510 --> 00:06:57.450 I feel I must compliment Mr. Reeves on his given name, it’s excellent. 00:06:57.450 --> 00:06:58.450 Very good. 00:06:58.450 --> 00:07:03.570 He first devised this digital method of transmitting and receiving voice communication in 1937, 00:07:03.570 --> 00:07:06.890 though it required extremely complex circuitry for the time. 00:07:06.890 --> 00:07:11.910 However, PCM transmission was used during World War 2 as a way to encrypt extremely 00:07:11.910 --> 00:07:15.394 important voice conversations, such as those between Winston Churchill 00:07:15.394 --> 00:07:17.620 and Franklin Delano Roosevelt. 00:07:17.620 --> 00:07:19.800 This encryption system was called SIGSALY, 00:07:19.800 --> 00:07:20.880 “SIGSALLY”? 00:07:20.880 --> 00:07:22.020 “SIGSALIE”? 00:07:22.460 --> 00:07:26.980 Or Project X, X System, Ciphony 1, or Green Hornet. 00:07:26.990 --> 00:07:31.550 Anyway, Project Green Sally X System Hornet 1 was much more complicated than simple Pulse 00:07:31.550 --> 00:07:35.450 Code Modulation, but PCM was a large part of its encryption. 00:07:35.450 --> 00:07:37.370 So how does PCM work? 00:07:37.370 --> 00:07:39.639 It’s actually simpler than it might seem at first. 00:07:39.639 --> 00:07:43.980 It’s rather like a system for repeatedly asking what the instantaneous amplitude of 00:07:43.980 --> 00:07:48.980 a signal is many thousands of times per second, then simply writing that down. 00:07:48.980 --> 00:07:51.200 Let’s look at a simple sine wave. 00:07:51.200 --> 00:07:55.450 If this were to be encoded on a vinyl record, the groove of the record would start out straight 00:07:55.450 --> 00:07:59.240 in the center, then move to the left as the signal intensity reached peak, then it would 00:07:59.240 --> 00:08:02.530 start to move to the right, keep moving, keep moving, and then it would pull back to the 00:08:02.530 --> 00:08:03.670 center. 00:08:03.670 --> 00:08:07.639 When it’s played back, the movement of the stylus as the walls of the groove wiggle it 00:08:07.639 --> 00:08:10.640 back and forth will recreate this signal. 00:08:10.640 --> 00:08:14.800 And audio tape does the same thing, except the intensity isn’t recorded as a physical 00:08:14.800 --> 00:08:18.080 movement, but as a degree of magnetization on the tape. 00:08:18.080 --> 00:08:21.650 But with PCM, we aren’t even trying to recreate the wave. 00:08:21.650 --> 00:08:24.980 Instead, we want to quantify it and play connect-the-dots. 00:08:24.980 --> 00:08:28.320 Let’s say I want to take 20 samples of this waveform. 00:08:28.320 --> 00:08:31.140 OK, I’ll divide it up into 20 chunks. 00:08:31.140 --> 00:08:34.820 Now I just need to define the detail I can have within each sample. 00:08:34.820 --> 00:08:37.040 Let’s put this on a scale of 0 to 15. 00:08:37.040 --> 00:08:38.940 That's 4 bits of resolution. 00:08:38.940 --> 00:08:42.890 Now, at each sampling point, we can take the closest value. 00:08:42.890 --> 00:08:46.920 This sine wave can now be represented as the following string of numbers. 00:08:46.920 --> 00:08:50.370 To get the sine wave back, we simply plot those numbers on a graph. 00:08:50.370 --> 00:08:52.520 Then, connect the dots. 00:08:52.520 --> 00:08:54.500 Tada! A sine… 00:08:54.500 --> 00:08:55.080 wave? 00:08:55.500 --> 00:08:56.980 Well, a sloppy sine wave. 00:08:56.990 --> 00:08:59.360 But that’s only because we weren’t very specific. 00:08:59.360 --> 00:09:03.770 We only took 20 samples, and each one could only be one of 16 values. 00:09:03.770 --> 00:09:07.850 But now we know the two most crucial parts of digital sound--the sample rate and the 00:09:07.850 --> 00:09:09.270 bit depth. 00:09:09.270 --> 00:09:13.540 Perhaps the most common sample rate and bit depth of digital sound is 44.1 kilohertz, 00:09:13.540 --> 00:09:15.000 16 bits. 00:09:15.000 --> 00:09:21.660 This means that every second, 44,100 samples are taken, and each sample can be one of 65,536 00:09:21.660 --> 00:09:24.720 values, or 2 to the power of 16. 00:09:24.720 --> 00:09:29.360 And that’s how devices like this, a Tascam DR-05, record sound. 00:09:29.360 --> 00:09:34.240 It’s looking at the voltage coming from the microphone, and taking precise measurements. 00:09:34.240 --> 00:09:40.610 Every 44.1 thousandth of a second, it takes a voltage reading, and, well, writes it down. 00:09:40.610 --> 00:09:45.870 It’s furiously quantifying and logging the voltage it measures with 16 bits of accuracy, 00:09:45.870 --> 00:09:49.970 and the result is a string of numbers that logically represent the shape of the sound 00:09:49.970 --> 00:09:53.430 waves that exerted pressure on the microphone’s diagram. 00:09:53.430 --> 00:09:54.610 Pretty neat, huh? 00:09:54.610 --> 00:09:58.470 And it can actually write down two numbers at a time, since this has two microphones 00:09:58.470 --> 00:10:00.180 and records in stereo. 00:10:00.180 --> 00:10:04.780 Inside this recorder is what’s called an analog-to-digital converter, or ADC. 00:10:04.780 --> 00:10:09.450 The “ADC” is the actual device responsible for creating the stream of samples. 00:10:09.450 --> 00:10:13.220 It takes the analog signal coming from the microphones themselves and converts it into 00:10:13.220 --> 00:10:14.930 a stream of discrete numbers. 00:10:14.930 --> 00:10:19.430 If you open the files it makes in audacity, you see what looks like a waveform of the sound. 00:10:19.430 --> 00:10:23.730 It is a waveform, but a waveform that’s been plotted precisely on a graph. 00:10:23.730 --> 00:10:25.872 Zoom way, way, way in on the waveform, 00:10:25.872 --> 00:10:29.310 and eventually you can see the individual samples themselves. 00:10:29.310 --> 00:10:31.220 And that’s all digital sound is-- 00:10:31.220 --> 00:10:33.960 it’s a huge list of numbers strung together in order. 00:10:33.960 --> 00:10:38.640 To get these numbers back into sound we can hear, we need to use the opposite of an analog-to-digital 00:10:38.640 --> 00:10:40.020 converter, or “ADC”. 00:10:40.029 --> 00:10:43.470 So, we’ll use a DAC, or Digital-to-analog converter. 00:10:43.470 --> 00:10:46.130 I like it when names make sense. 00:10:46.130 --> 00:10:50.180 A DAC will read the string of numbers, and generate an analog voltage based upon their 00:10:50.180 --> 00:10:51.320 values. 00:10:51.320 --> 00:10:55.150 The DAC will smooth out the choppiness of the samples a bit to make the resulting sound 00:10:55.150 --> 00:10:59.970 a little more natural, and now you’ve got an analog signal to send into an amplifier 00:10:59.970 --> 00:11:01.920 and drive a loudspeaker. 00:11:01.920 --> 00:11:06.170 The result is a near-perfect reproduction of the originally recorded sound. 00:11:06.170 --> 00:11:11.029 Here’s a very crude analogy to explain the difference between analog and digital sound. 00:11:11.029 --> 00:11:16.000 A vinyl record’s walls generate an analog signal by moving the stylus left and right... 00:11:16.000 --> 00:11:17.760 as well as up and down. 00:11:17.760 --> 00:11:21.250 It’s diagonally moved for stereo, but just imagine for a moment that it’s just left 00:11:21.250 --> 00:11:22.250 and right. 00:11:22.250 --> 00:11:26.540 A record directly creates the analog signal via the motion of the stylus. 00:11:26.540 --> 00:11:32.210 But a digital sound source is instead sort of like a virtual stylus riding in a virtual groove. 00:11:32.210 --> 00:11:36.660 The sound samples are snapshots in time of where the stylus was. 00:11:36.660 --> 00:11:41.820 A DAC will then create an analog signal by running a virtual stylus through this virtual 00:11:41.820 --> 00:11:46.490 groove and placing it at exactly the correct location--and thus generating the appropriate 00:11:46.490 --> 00:11:49.839 voltage level--as defined by the samples. 00:11:49.839 --> 00:11:54.040 By using a giant list of numbers to recreate sound, instead of the physical properties 00:11:54.040 --> 00:11:59.480 of a plastic disc, the sound can be reproduced flawlessly and accurately with no reliance 00:11:59.480 --> 00:12:03.670 on the record player’s cartridge properties, the integrity of its stylus, it’s motor, 00:12:03.670 --> 00:12:05.800 the quality of the vinyl etc. 00:12:05.800 --> 00:12:10.220 The biggest boon of digital sound was that it eliminated all of the little nuances that 00:12:10.220 --> 00:12:13.180 might change how a recording sounds. 00:12:13.180 --> 00:12:15.800 Digital sound is in a sense, absolute. 00:12:15.800 --> 00:12:20.730 But getting digital sound into the hands of the average consumer took a long while. 00:12:20.730 --> 00:12:25.700 DACs and “ADCs” were expensive components, and the amount of raw data generated by sound 00:12:25.700 --> 00:12:28.820 recording was immense for the standards of the time. 00:12:28.820 --> 00:12:33.810 Although 650 megabytes, the data equivalent of the first compact discs, is a paltry sum 00:12:33.810 --> 00:12:39.630 of data in the 21st century, it was unimaginably huge in the early 1970’s, when the first 00:12:39.630 --> 00:12:42.370 commercial digital sound recording took place. 00:12:42.370 --> 00:12:48.270 For context, the Commodore 64, released the same year as the compact disc, has 64 kilobytes 00:12:48.270 --> 00:12:51.850 of ram, and that was considered huge for the time. 00:12:51.850 --> 00:12:55.930 A compact disc held roughly ten thousands times as much data. 00:12:55.930 --> 00:13:00.080 64 kilobtyes of CD quality audio lasts this long; 00:13:00.300 --> 00:13:00.800 (clip) 00:13:01.120 --> 00:13:02.620 That’s not super helpful. 00:13:02.620 --> 00:13:06.600 When we continue, we’ll look at the methods that were used to store data from digital 00:13:06.600 --> 00:13:10.840 recordings, and we’ll discuss the rise of the compact disc as a robust, consumer-friendly 00:13:10.840 --> 00:13:14.020 format for digital sound reproduction and distribution. 00:13:14.020 --> 00:13:16.690 Thanks for watching, I hope you enjoyed the video! 00:13:16.690 --> 00:13:20.170 If this is your first time coming across the channel and you liked what you saw, please 00:13:20.170 --> 00:13:22.300 consider subscribing to Technology Connections. 00:13:22.300 --> 00:13:27.090 Don’t forget you can also follow me on Twitter @TechConnectify, and you might enjoy the second 00:13:27.090 --> 00:13:31.790 channel, Technology Connection 2, where I talk about stuff and don’t prepare for anything. 00:13:31.790 --> 00:13:37.279 Also, thanks to Lord Telaneo on Twitter, there is also a Technology Connections Subreddit. 00:13:37.279 --> 00:13:42.770 I really don’t know reddit at all, but you will also find me there as TechConnectify. 00:13:42.770 --> 00:13:46.870 As always, thank you to everyone who supports this channel on Patreon, especially the wonderful 00:13:46.870 --> 00:13:49.320 folks that have been scrolling up your screen. 00:13:49.320 --> 00:13:52.700 It is with the support of people like you that I’m able to make these videos. 00:13:52.700 --> 00:13:53.899 Thank you. 00:13:53.899 --> 00:13:57.060 If you’d like to you join these awesome people and support the channel too, why not 00:13:57.060 --> 00:13:58.830 take a look at my Patreon page. 00:13:58.830 --> 00:14:01.600 Thank you for your consideration, and I’ll see you next time!