WEBVTT Kind: captions Language: en 00:00:00.049 --> 00:00:05.060 If you were born some time after, oh, let’s say 1988, then when you see this symbol, the 00:00:05.060 --> 00:00:10.260 Dolby logo, you probably think of a movie theater or DVD with Dolby Digital 7.1 surround 00:00:10.260 --> 00:00:14.519 sound, or perhaps Dolby’s newer technologies such as Atmos or Dolby Vision HDR. 00:00:14.519 --> 00:00:18.480 So you might be a little surprised to see it on an old cassette tape. 00:00:18.480 --> 00:00:19.920 What’s it doing there? 00:00:19.920 --> 00:00:21.919 Well, Dolby didn’t start out in the movie business. 00:00:21.919 --> 00:00:25.899 In fact, they started in the business of sound. 00:00:25.899 --> 00:00:30.449 I suppose that’s not a surprise, after all that’s pretty much exactly what they do 00:00:30.449 --> 00:00:34.280 today, but their first product had nothing to do with surround sound, movies, or anything 00:00:34.280 --> 00:00:36.010 high-tech like that. 00:00:36.010 --> 00:00:38.670 Their first product was Dolby Noise Reduction. 00:00:38.670 --> 00:00:44.219 Ray Dolby started the company in 1965, and their first product was the Dolby 301 unit, 00:00:44.219 --> 00:00:47.289 a machine designed to help eliminate tape noise. 00:00:47.289 --> 00:00:51.500 With magnetic tape, the random structure of the particles that make up the tape create 00:00:51.500 --> 00:00:56.030 an audible signal that we don’t want, a quiet but still noticeable noise that sounds 00:00:56.030 --> 00:00:57.310 like a hiss. 00:00:57.310 --> 00:01:00.800 This noise is always there, again due to the structure of the tape itself. 00:01:00.800 --> 00:01:04.560 During loud sections of a recording, it’s not that noticeable because compared the the 00:01:04.560 --> 00:01:06.500 signal, it’s barely perceptible. 00:01:06.500 --> 00:01:10.380 But during quiet parts of a song, it becomes much more of a problem. 00:01:10.380 --> 00:01:13.680 Through the magic of video editing, I’m going to take the audio from this video and 00:01:13.680 --> 00:01:16.740 slap it on a cassette so you hear what this noise sounds like. 00:01:17.740 --> 00:01:18.740 There. 00:01:18.740 --> 00:01:19.740 That’s, better? 00:01:19.740 --> 00:01:22.930 Well I suppose it’s actually worse, but there you go, that’s the problem. 00:01:22.930 --> 00:01:27.690 This tape noise permeates all recordings, and particularly in quiet or silent portions, 00:01:27.690 --> 00:01:29.450 it’s quite noticeable. 00:01:29.450 --> 00:01:32.000 Ray Dolby wanted to figure out a way to get rid of it. 00:01:32.000 --> 00:01:33.650 His solution is quite elegant. 00:01:33.650 --> 00:01:36.830 First, let me turn on Dolby Noise Reduction so you can hear the difference. 00:01:37.830 --> 00:01:38.880 There, that is better. 00:01:38.880 --> 00:01:40.540 Now the noise is much less noticeable. 00:01:40.540 --> 00:01:43.750 It’s still there, sure, but it’s definitely not as annoying. 00:01:43.750 --> 00:01:47.430 But what’s most interesting is the fact that the actual signal we wanted is unaffected. 00:01:47.430 --> 00:01:51.190 I don’t sound any different now that we’re using noise reduction, the only thing that 00:01:51.190 --> 00:01:54.680 happened is now you can hear me better because the noise has gone away. 00:01:54.680 --> 00:01:55.830 So how does this work? 00:01:55.830 --> 00:01:59.660 Well, the first thing Dolby did was to figure out what the tape noise was. 00:01:59.660 --> 00:02:04.410 By using spectral analysis of a blank tape, you can see what sound frequencies make up 00:02:04.410 --> 00:02:05.410 the noise. 00:02:05.410 --> 00:02:08.090 Therefore, you can get a noise profile. 00:02:08.090 --> 00:02:11.900 Once Dolby had figured out the particular frequencies that made up the sound, it was 00:02:11.900 --> 00:02:14.730 simply a matter of using an equalizer to tune out those frequencies. 00:02:14.730 --> 00:02:18.740 You’ve probably used an equalizer before as they come built into many music playing 00:02:18.740 --> 00:02:21.620 applications such as iTunes and the like. 00:02:21.620 --> 00:02:25.780 The equalizer has a slider for particular ranges of sound frequencies, and you can either 00:02:25.780 --> 00:02:28.260 boost each band or suppress it. 00:02:28.260 --> 00:02:32.030 Dolby noise reduction is a lot like taking the frequencies that make up noise and tuning 00:02:32.030 --> 00:02:34.650 them down a lot to make the noise quieter. 00:02:34.650 --> 00:02:36.150 OK, so let’s do that. 00:02:36.150 --> 00:02:40.690 (pause) Now the noise is a lot less noticeable, But we have a new problem. 00:02:40.690 --> 00:02:44.790 Now the sound we want is dull, and my voice sounds muffled. 00:02:44.790 --> 00:02:48.000 This isn’t really any better, it’s just a different problem. 00:02:48.000 --> 00:02:50.520 So how come nothing changed when we used it the first time? 00:02:50.520 --> 00:02:52.590 Let me turn that back off. 00:02:52.590 --> 00:02:57.090 Nothing changed the first time because Dolby Noise Reduction is actually a two way process. 00:02:57.090 --> 00:03:02.450 In order for it to work, the recording also has to be processed with Dolby NR. 00:03:02.450 --> 00:03:05.990 During recording, the Dolby Chip inside the cassette deck actually does the opposite of 00:03:05.990 --> 00:03:07.910 what it does during playback. 00:03:07.910 --> 00:03:12.020 Instead of cutting the frequencies associated with noise, it boosts them. 00:03:12.020 --> 00:03:15.420 Now I’ll turn on Dolby noise reduction at the time of recording, but I’ll play the 00:03:15.420 --> 00:03:17.270 tape back without Dolby turned on. 00:03:17.270 --> 00:03:18.270 (pause) 00:03:18.270 --> 00:03:22.730 Now my voice sounds really harsh, with ess sounds being particularly annoying. 00:03:22.730 --> 00:03:28.260 The ess sound is so satisfactorily nasty such that some sadistic, senseless sad sack might 00:03:28.260 --> 00:03:32.470 succumb to the sickening task of overusing these sounds to facilitate the driving of 00:03:32.470 --> 00:03:33.680 a point home. 00:03:33.680 --> 00:03:36.220 (clears throat) Yes, but there is a reason for this. 00:03:36.220 --> 00:03:39.990 By over-boosting these frequencies during recording, we eliminate the problem of a dull 00:03:39.990 --> 00:03:41.690 recording when using noise reduction. 00:03:41.690 --> 00:03:44.020 Let’s look at the equalizer again. 00:03:44.020 --> 00:03:46.910 This is what happens when we use dolby during playback. 00:03:46.910 --> 00:03:49.660 These frequencies are pulled down to kill the noise. 00:03:49.660 --> 00:03:53.930 But by doing this, we also pull down any components of the signal , that is the sounds that we 00:03:53.930 --> 00:03:55.599 want, that lie within that range. 00:03:55.599 --> 00:03:59.790 But if you boost the frequencies that lie in the spectrum with the noise while recording, 00:03:59.790 --> 00:04:04.020 you end up with a recording that has these particular frequencies recorded too loud. 00:04:04.020 --> 00:04:08.400 Then, when you pull those frequencies down during playback, you not only pull the noise 00:04:08.400 --> 00:04:12.680 down and make it quieter, but you also bring the boosted parts of the signal down back 00:04:12.680 --> 00:04:13.880 to where they should be. 00:04:13.880 --> 00:04:18.579 In essence, the boosting of sounds we want during recording is canceled out bringing 00:04:18.579 --> 00:04:21.109 them back to normal during playback. 00:04:21.109 --> 00:04:24.300 Because any components of the signal that lie within the noise profile are boosted to 00:04:24.300 --> 00:04:28.849 compensate for the noise reduction, the effect is that only the noise is reduced. 00:04:28.849 --> 00:04:30.759 The signal remains intact. 00:04:30.759 --> 00:04:32.860 This is a process known as companding. 00:04:32.860 --> 00:04:36.860 The dynamic range of the recording, that is the difference between loud and soft sounds, 00:04:36.860 --> 00:04:40.550 it compressed during recording because some frequencies are recorded louder than they 00:04:40.550 --> 00:04:41.949 should be. 00:04:41.949 --> 00:04:46.330 Then during playback, the dynamic range is expanded by lowering those frequencies, and 00:04:46.330 --> 00:04:47.840 thus the noise floor. 00:04:47.840 --> 00:04:49.309 There is a catch, though. 00:04:49.309 --> 00:04:53.430 Because some sounds are boosted by the dolby circuitry when recording, the record levels 00:04:53.430 --> 00:04:57.689 that you use must be reduced a little bit to ensure they don’t get distorted. 00:04:57.689 --> 00:05:02.199 That’s why the Dolby symbol usually appears on the level meter of a tape deck--this is 00:05:02.199 --> 00:05:06.150 the point on the meter that should not be exceeded when using noise reduction. 00:05:06.150 --> 00:05:09.479 When setting the level for recording, you’d want the meter to just barely touch it at 00:05:09.479 --> 00:05:10.750 the loudest sections. 00:05:10.750 --> 00:05:15.639 The Dolby system used on consumer machines like this wasn’t the best Dolby had to offer. 00:05:15.639 --> 00:05:19.789 Dolby A noise reduction is what Ray Dolby first invented, and was what was found in 00:05:19.789 --> 00:05:21.979 the professional recording studio. 00:05:21.979 --> 00:05:26.469 But Dolby-B, what’s used on the vast majority of cassettes, worked pretty well, and it was 00:05:26.469 --> 00:05:29.099 a lot simpler and thus cheaper to implement. 00:05:29.099 --> 00:05:34.159 Later consumer systems, such as Dolby C and Dolby S, further improved on the noise reduction 00:05:34.159 --> 00:05:35.159 process. 00:05:35.159 --> 00:05:40.300 Dolby C came about in 1980, and it worked very well, about 60 % better than B. The problem 00:05:40.300 --> 00:05:45.910 with Dolby C, however, was that on non-compatible equipment, the tape sounds really weird. 00:05:45.910 --> 00:05:46.910 Have a listen. 00:05:46.910 --> 00:06:14.809 Here’s a clip of some music from the YouTube Audio Library. 00:06:14.809 --> 00:06:25.389 Now here’s what it sounds like on a Dolby-C encoded tape, played back without Dolby circuitry. 00:06:25.389 --> 00:06:45.029 That’s messed up. 00:06:45.029 --> 00:06:48.309 Turning on Dolby B helped a little, but not much. 00:06:48.309 --> 00:06:52.389 Dolby C used a much more complicated process to reduce noise, and that’s why recordings 00:06:52.389 --> 00:06:54.029 made with it are less compatible. 00:06:54.029 --> 00:06:58.499 It also meant that pre-recorded tapes didn’t shift to Dolby C, since no one wants to buy 00:06:58.499 --> 00:07:02.119 music that doesn’t sound any good, and even fewer people were willing to put down the 00:07:02.119 --> 00:07:03.749 money for a new cassette deck. 00:07:03.749 --> 00:07:08.360 The next consumer development, Dolby S, was supposedly so good that with a decent quality 00:07:08.360 --> 00:07:12.389 Chromium tape, you couldn’t tell the difference between a cassette encoded with Dolby S and 00:07:12.389 --> 00:07:13.680 a CD. 00:07:13.680 --> 00:07:18.080 It also was more compatible with Dolby B equipment, not suffering from the distortion of dolby 00:07:18.080 --> 00:07:23.611 C. The problem with Dolby S, though, was that it came about far too late, in 1989, with 00:07:23.611 --> 00:07:26.409 the CD already having found a prominent place in the market. 00:07:26.409 --> 00:07:30.469 The best noise-reduction technologies, such as those used in the professional Dolby SR 00:07:30.469 --> 00:07:35.330 or the competing DBX system, used very aggressive companding. 00:07:35.330 --> 00:07:39.349 The idea is to keep the signal recorded on the tape loud relative to the noise at all 00:07:39.349 --> 00:07:40.349 times. 00:07:40.349 --> 00:07:44.159 Remember, in loud sections of the recording, the noise isn’t noticeable compared to the 00:07:44.159 --> 00:07:45.159 signal. 00:07:45.159 --> 00:07:49.639 So the best possible system would make the recording loud all the time, and during playback 00:07:49.639 --> 00:07:54.189 could tell which parts were intended to be quiet and thus lower the volume of the whole 00:07:54.189 --> 00:07:59.050 signal back to normal, therefore making the noise practically imperceptible. 00:07:59.050 --> 00:08:02.860 Though this isn’t quite how companding is accomplished in the best systems, the effect 00:08:02.860 --> 00:08:03.940 is similar. 00:08:03.940 --> 00:08:08.050 But the problem with using this technique is that listening to a DBX encoded tape without 00:08:08.050 --> 00:08:11.729 a DBX capable decoder is very unpleasant. 00:08:11.729 --> 00:08:13.099 The same goes for Dolby SR. 00:08:13.099 --> 00:08:17.409 And that’s precisely why the better systems never took off in the consumer space. 00:08:17.409 --> 00:08:21.879 Dolby B was simple and still pretty effective, but best of all a Dolby-B encoded tape could 00:08:21.879 --> 00:08:25.979 be played back on a standard cassette deck with reasonable results, again with the only 00:08:25.979 --> 00:08:29.389 difference being an over-emphasis of the treble sounds. 00:08:29.389 --> 00:08:33.039 And with most stereos having a bass and treble adjustment, just slightly turning down the 00:08:33.039 --> 00:08:36.839 treble would bring the sound back to more or less normal while still providing some 00:08:36.839 --> 00:08:41.199 noise reduction, of course the noise reduction you’d get would be less precise and effective. 00:08:41.199 --> 00:08:45.350 We don’t use noise reduction any more because, well, noise isn’t a problem any more. 00:08:45.350 --> 00:08:48.899 Now that we use digital sound in virtually all applications, there’s no need to add 00:08:48.899 --> 00:08:51.019 this sort of complex circuitry. 00:08:51.019 --> 00:08:53.630 But I’m still impressed by the elegant solution to the problem. 00:08:53.630 --> 00:08:58.209 Automatic pre-emphasis of sounds within the noise profile when recording and then de-emphasis 00:08:58.209 --> 00:09:03.210 when playing it back is so delightfully simple, effective, and logical, that I can’t help 00:09:03.210 --> 00:09:04.220 but admire it. 00:09:04.220 --> 00:09:05.490 Thanks for watching. 00:09:05.490 --> 00:09:10.560 If you like this sort of video, be sure to give it a thumbs up and subscribe to Technology 00:09:10.560 --> 00:09:11.560 Connections. 00:09:11.560 --> 00:09:14.579 I’m doing my best to keep videos like this coming your way. 00:09:14.579 --> 00:09:17.500 If you have ideas of what to explore next, leave them in the comments. 00:09:17.500 --> 00:09:18.500 I’d love to hear them. 00:09:18.500 --> 00:09:19.050 I’ll see you next time!