WEBVTT 00:00:00.000 --> 00:00:02.455 align:middle line:90% [SQUEAKING] 00:00:02.455 --> 00:00:04.419 align:middle line:90% [RUSTLING] 00:00:04.419 --> 00:00:07.365 align:middle line:90% [CLICKING] 00:00:07.365 --> 00:00:10.270 align:middle line:90% 00:00:10.270 --> 00:00:12.020 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: Today, we're 00:00:12.020 --> 00:00:16.490 align:middle line:84% going to do two really important things that I really 00:00:16.490 --> 00:00:19.490 align:middle line:84% want to get through today, and the first one 00:00:19.490 --> 00:00:24.450 align:middle line:84% involves the final project for the class. 00:00:24.450 --> 00:00:28.950 align:middle line:84% So what I'm going to ask you to do is to take a minute, 00:00:28.950 --> 00:00:31.970 align:middle line:84% and after I tell you what you're going to be doing, 00:00:31.970 --> 00:00:34.650 align:middle line:90% find a partner. 00:00:34.650 --> 00:00:39.690 align:middle line:84% If we're not and even number, you can do one group of three. 00:00:39.690 --> 00:00:44.390 align:middle line:84% But I prefer groups of two have tended to work a lot better, 00:00:44.390 --> 00:00:48.200 align:middle line:84% that you're going to be willing to work on a project starting 00:00:48.200 --> 00:00:51.690 align:middle line:84% next week and start thinking about what you would like to do. 00:00:51.690 --> 00:00:56.570 align:middle line:84% It can be anything related to music theory and analysis 00:00:56.570 --> 00:00:59.590 align:middle line:84% in the symbolic domain, that is to say, 00:00:59.590 --> 00:01:03.220 align:middle line:84% working with notes and scores, not primarily working 00:01:03.220 --> 00:01:06.550 align:middle line:90% with audio or things like that. 00:01:06.550 --> 00:01:10.780 align:middle line:84% Would you want to do that, take Eran Egozy's amazing classes 00:01:10.780 --> 00:01:14.260 align:middle line:84% where you can do great projects on audio. 00:01:14.260 --> 00:01:17.810 align:middle line:84% And it can be a research type problem. 00:01:17.810 --> 00:01:20.850 align:middle line:84% You can try to solve a research problem. 00:01:20.850 --> 00:01:23.950 align:middle line:84% It can be more of a business development and entrepreneurship 00:01:23.950 --> 00:01:28.330 align:middle line:84% type problem, trying to build a product that does something 00:01:28.330 --> 00:01:32.830 align:middle line:84% interesting with music theory, or it can be a creative project. 00:01:32.830 --> 00:01:34.970 align:middle line:84% I mentioned at the beginning of the semester, 00:01:34.970 --> 00:01:38.440 align:middle line:84% but I want to say it again, the only restriction 00:01:38.440 --> 00:01:41.150 align:middle line:84% besides primarily symbolically based 00:01:41.150 --> 00:01:44.390 align:middle line:84% is that, if you are doing a creative project, 00:01:44.390 --> 00:01:48.072 align:middle line:84% I don't want it to be based on deep learning AI. 00:01:48.072 --> 00:01:52.150 align:middle line:84% And the reason for that is as a safety for y'all. 00:01:52.150 --> 00:01:56.440 align:middle line:84% I've allowed it in the past, and it hasn't worked very well. 00:01:56.440 --> 00:02:02.980 align:middle line:84% And we will do a lecture on AI later on, or a topic on AI, 00:02:02.980 --> 00:02:05.310 align:middle line:84% and then I'll be able to explain a little bit more 00:02:05.310 --> 00:02:08.940 align:middle line:84% on what the constraint that tends to make those 00:02:08.940 --> 00:02:11.700 align:middle line:90% projects not work very well. 00:02:11.700 --> 00:02:14.460 align:middle line:84% So has anybody already been thinking 00:02:14.460 --> 00:02:16.293 align:middle line:84% over the course of the semester of something 00:02:16.293 --> 00:02:17.377 align:middle line:90% that you're interested in? 00:02:17.377 --> 00:02:19.830 align:middle line:84% I'm not going to ask you to say it, but I'm just curious. 00:02:19.830 --> 00:02:22.570 align:middle line:84% Who's been thinking a little bit, a little bit, a little bit? 00:02:22.570 --> 00:02:25.110 align:middle line:90% OK, so some over here. 00:02:25.110 --> 00:02:29.580 align:middle line:84% So let's take 10 minutes, five minutes 00:02:29.580 --> 00:02:31.960 align:middle line:84% to find somebody that you'd like to work with. 00:02:31.960 --> 00:02:34.290 align:middle line:90% So let's all stand up. 00:02:34.290 --> 00:02:35.630 align:middle line:90% OK we'll get to-- 00:02:35.630 --> 00:02:39.040 align:middle line:90% 00:02:39.040 --> 00:02:41.680 align:middle line:84% no, actually, your computers are mostly closed. 00:02:41.680 --> 00:02:44.580 align:middle line:84% So let me just remind us a little bit 00:02:44.580 --> 00:02:48.357 align:middle line:84% of the partitive topic for the week and talk about some 00:02:48.357 --> 00:02:48.940 align:middle line:90% of the things. 00:02:48.940 --> 00:02:50.470 align:middle line:90% Yes, yes, it's down. 00:02:50.470 --> 00:02:51.520 align:middle line:90% Thank you. 00:02:51.520 --> 00:02:54.180 align:middle line:90% Just shout when you get that. 00:02:54.180 --> 00:02:57.440 align:middle line:84% I did remember to turn the signal back on, but yep. 00:02:57.440 --> 00:03:00.230 align:middle line:90% Thank you. 00:03:00.230 --> 00:03:04.910 align:middle line:84% So we're going to talk about some of the other things 00:03:04.910 --> 00:03:06.680 align:middle line:84% that we were talking about last class 00:03:06.680 --> 00:03:08.310 align:middle line:90% with algorithmic composition. 00:03:08.310 --> 00:03:11.750 align:middle line:84% So first off, who now has been able to access 00:03:11.750 --> 00:03:14.390 align:middle line:90% the ILLIAC suite and recording? 00:03:14.390 --> 00:03:15.450 align:middle line:90% OK, great, super. 00:03:15.450 --> 00:03:19.010 align:middle line:84% So we're in a better position there. 00:03:19.010 --> 00:03:25.820 align:middle line:84% We talked about whether or not the ILLIAC suite was more 00:03:25.820 --> 00:03:29.070 align:middle line:84% written to create music that sounded like a human, 00:03:29.070 --> 00:03:33.090 align:middle line:84% or whether it was created to do something very new. 00:03:33.090 --> 00:03:35.550 align:middle line:84% And what did we come to this consensus? 00:03:35.550 --> 00:03:39.260 align:middle line:90% Anyone want to throw that out? 00:03:39.260 --> 00:03:40.100 align:middle line:90% Or what did you-- 00:03:40.100 --> 00:03:41.610 align:middle line:84% AUDIENCE: 30 towards seeing something new? 00:03:41.610 --> 00:03:42.290 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: What's that? 00:03:42.290 --> 00:03:44.220 align:middle line:84% 70/30 toward doing something new. 00:03:44.220 --> 00:03:45.690 align:middle line:90% Yeah, yeah, something like that. 00:03:45.690 --> 00:03:49.340 align:middle line:84% Some percentage, some mixture of some parts of what 00:03:49.340 --> 00:03:52.380 align:middle line:84% they were trying to do really was about, hey, 00:03:52.380 --> 00:03:53.880 align:middle line:90% this is something humans do. 00:03:53.880 --> 00:03:55.880 align:middle line:84% Let's see if it can computer can do it. 00:03:55.880 --> 00:03:58.780 align:middle line:84% And then there was some moments, like the one 00:03:58.780 --> 00:04:00.760 align:middle line:84% with the algorithmic pudding of claps 00:04:00.760 --> 00:04:03.880 align:middle line:84% and hitting the violins and stuff like that 00:04:03.880 --> 00:04:08.190 align:middle line:84% at different points, which was a little bit more computery. 00:04:08.190 --> 00:04:12.590 align:middle line:84% We also talked about Markov chains in the last class, 00:04:12.590 --> 00:04:15.310 align:middle line:84% and we'll get a little bit later on into some 00:04:15.310 --> 00:04:18.430 align:middle line:84% of the more sophisticated statistical learning 00:04:18.430 --> 00:04:23.090 align:middle line:84% that people are doing today with AI learning. 00:04:23.090 --> 00:04:25.490 align:middle line:84% But a lot of it is very, very similar. 00:04:25.490 --> 00:04:29.810 align:middle line:84% It's just a lot faster, a lot bigger than things. 00:04:29.810 --> 00:04:32.620 align:middle line:84% But I want to talk about some things, 00:04:32.620 --> 00:04:35.285 align:middle line:90% and we'll to get to that later. 00:04:35.285 --> 00:04:36.410 align:middle line:90% Pretend I didn't show that. 00:04:36.410 --> 00:04:42.100 align:middle line:84% So there's two ways that we can distinguish 00:04:42.100 --> 00:04:43.610 align:middle line:90% algorithmic composition. 00:04:43.610 --> 00:04:46.210 align:middle line:84% One of them is whether it's deterministic 00:04:46.210 --> 00:04:50.590 align:middle line:84% or if it has a random chance in it, whether or not every time 00:04:50.590 --> 00:04:53.430 align:middle line:84% you play the piece or run the piece, 00:04:53.430 --> 00:04:56.840 align:middle line:84% it will be a different piece or it will be the same. 00:04:56.840 --> 00:04:58.560 align:middle line:90% So I want to start by giving-- 00:04:58.560 --> 00:05:02.450 align:middle line:84% I'm going to go pretty egotistically with two pieces 00:05:02.450 --> 00:05:03.360 align:middle line:90% that I've written. 00:05:03.360 --> 00:05:05.068 align:middle line:84% The first one is not even really a piece, 00:05:05.068 --> 00:05:06.840 align:middle line:84% but it's kind of a little demonstration. 00:05:06.840 --> 00:05:09.980 align:middle line:84% It starts with a demo that you might 00:05:09.980 --> 00:05:13.640 align:middle line:84% have seen in a science class or physics class, where 00:05:13.640 --> 00:05:17.700 align:middle line:84% all the balls are moving at different lengths, 00:05:17.700 --> 00:05:24.440 align:middle line:84% but they do some interesting optical illusion effects 00:05:24.440 --> 00:05:27.750 align:middle line:90% at various times. 00:05:27.750 --> 00:05:28.250 align:middle line:90% ASMR. 00:05:28.250 --> 00:05:43.993 align:middle line:90% 00:05:43.993 --> 00:05:46.410 align:middle line:84% AUDIENCE: Is there supposed to be music in the background? 00:05:46.410 --> 00:05:48.368 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: No, no music yet. 00:05:48.368 --> 00:06:01.400 align:middle line:90% 00:06:01.400 --> 00:06:04.010 align:middle line:84% OK, so I thought, what if there was 00:06:04.010 --> 00:06:05.910 align:middle line:90% a musical equivalent of this? 00:06:05.910 --> 00:06:08.030 align:middle line:84% What if we could write a computer program that 00:06:08.030 --> 00:06:13.070 align:middle line:84% had 12 lines, or eight lines, or something like that, that all 00:06:13.070 --> 00:06:16.340 align:middle line:84% started in phase, and then slowly got out of phase, 00:06:16.340 --> 00:06:18.630 align:middle line:84% and came back together like the pendulum? 00:06:18.630 --> 00:06:23.850 align:middle line:84% And so this is a little bit what I came up with. 00:06:23.850 --> 00:06:27.308 align:middle line:90% [MUSICAL NOTES] 00:06:27.308 --> 00:06:58.950 align:middle line:90% 00:06:58.950 --> 00:07:03.690 align:middle line:84% OK, I can't even follow my own score properly. 00:07:03.690 --> 00:07:05.935 align:middle line:90% So what's that? 00:07:05.935 --> 00:07:07.560 align:middle line:84% AUDIENCE: With the double dotted 16th-- 00:07:07.560 --> 00:07:09.727 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: 16th rest, yeah, yeah. 00:07:09.727 --> 00:07:12.430 align:middle line:84% So sometimes when you're making an algorithmic composition, 00:07:12.430 --> 00:07:16.410 align:middle line:84% maybe you want to think a little bit about the notation, 00:07:16.410 --> 00:07:18.300 align:middle line:84% or maybe even better just to hear it 00:07:18.300 --> 00:07:20.260 align:middle line:84% without seeing the notation at all. 00:07:20.260 --> 00:07:24.840 align:middle line:84% So one of the questions is, if the computer is just setting it 00:07:24.840 --> 00:07:28.860 align:middle line:84% up so that each note comes in at a slightly different time 00:07:28.860 --> 00:07:30.870 align:middle line:84% from the previous one, what kinds of things 00:07:30.870 --> 00:07:35.776 align:middle line:84% does the composer have for her, or his or their creativity? 00:07:35.776 --> 00:07:40.110 align:middle line:90% 00:07:40.110 --> 00:07:41.200 align:middle line:90% Yeah, go ahead Jason. 00:07:41.200 --> 00:07:43.140 align:middle line:84% AUDIENCE: Which notes, which pitches show up? 00:07:43.140 --> 00:07:44.740 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: So pitching which pitches show up. 00:07:44.740 --> 00:07:45.430 align:middle line:90% Good, good. 00:07:45.430 --> 00:07:48.020 align:middle line:90% Cause that could, in fact-- 00:07:48.020 --> 00:07:50.330 align:middle line:90% let me do it. 00:07:50.330 --> 00:07:52.870 align:middle line:90% Here is with, instead of using-- 00:07:52.870 --> 00:07:54.950 align:middle line:84% I think I used eight notes out of the octave. 00:07:54.950 --> 00:07:57.650 align:middle line:90% Here is all 12 notes. 00:07:57.650 --> 00:08:01.094 align:middle line:90% [MUSICAL NOTES] 00:08:01.094 --> 00:08:08.980 align:middle line:90% 00:08:08.980 --> 00:08:10.990 align:middle line:84% And that sounds a little bit more familiar 00:08:10.990 --> 00:08:13.510 align:middle line:90% and maybe a little bit closer. 00:08:13.510 --> 00:08:16.845 align:middle line:84% Why might I not have wanted to use this as my main piece? 00:08:16.845 --> 00:08:19.994 align:middle line:90% [MUSICAL NOTES] 00:08:19.994 --> 00:08:22.678 align:middle line:90% 00:08:22.678 --> 00:08:25.220 align:middle line:84% It has something to do with the duration of the piece for me, 00:08:25.220 --> 00:08:28.340 align:middle line:84% that it ended up being a little bit too long. 00:08:28.340 --> 00:08:30.860 align:middle line:84% So yeah, definitely what notes we use. 00:08:30.860 --> 00:08:34.045 align:middle line:90% [MUSICAL NOTES] 00:08:34.045 --> 00:08:34.960 align:middle line:90% 00:08:34.960 --> 00:08:39.100 align:middle line:84% We can even go to weird things that a human can't usually 00:08:39.100 --> 00:08:41.253 align:middle line:90% play on a piano. 00:08:41.253 --> 00:08:44.557 align:middle line:90% [MUSICAL NOTES] 00:08:44.557 --> 00:08:46.930 align:middle line:90% 00:08:46.930 --> 00:08:50.500 align:middle line:84% So this is 19 tone division of the octave. 00:08:50.500 --> 00:08:53.160 align:middle line:84% Let's you use a few more notes per octave. 00:08:53.160 --> 00:08:55.780 align:middle line:90% [MUSICAL NOTES] 00:08:55.780 --> 00:08:57.400 align:middle line:90% Oh, blanking. 00:08:57.400 --> 00:08:58.820 align:middle line:90% It stops the music. 00:08:58.820 --> 00:08:59.320 align:middle line:90% Good. 00:08:59.320 --> 00:08:59.820 align:middle line:90% Yeah. 00:08:59.820 --> 00:09:01.100 align:middle line:90% AUDIENCE: How do you go to 19? 00:09:01.100 --> 00:09:02.810 align:middle line:84% I would think it would be 24 or something? 00:09:02.810 --> 00:09:04.268 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: Yeah. 00:09:04.268 --> 00:09:06.610 align:middle line:84% Sorry, it's a little digression, but it's kind of cool 00:09:06.610 --> 00:09:08.180 align:middle line:90% if anybody's interested in it. 00:09:08.180 --> 00:09:15.260 align:middle line:84% A 19 tone division of the octave has two pretty cool properties. 00:09:15.260 --> 00:09:20.470 align:middle line:84% One is that it has major and minor triads that are very, very 00:09:20.470 --> 00:09:23.110 align:middle line:84% close to acoustically-- or major and minor thirds 00:09:23.110 --> 00:09:26.510 align:middle line:84% very close to acoustically correct, 00:09:26.510 --> 00:09:29.860 align:middle line:84% while also having perfect fourths and perfect fifths 00:09:29.860 --> 00:09:30.740 align:middle line:90% are pretty good. 00:09:30.740 --> 00:09:32.510 align:middle line:90% So that's one reason. 00:09:32.510 --> 00:09:36.310 align:middle line:84% The other reason for it is a lot of 19 tone piano-- 00:09:36.310 --> 00:09:39.850 align:middle line:84% a lot of 19 tone music is written for two piano duets 00:09:39.850 --> 00:09:46.790 align:middle line:84% where you keep the five black keys the same between each 00:09:46.790 --> 00:09:50.870 align:middle line:84% of the pianos, and then you retune the white keys 00:09:50.870 --> 00:09:55.927 align:middle line:84% so that you have a separate note for each of them. 00:09:55.927 --> 00:09:56.510 align:middle line:90% Is that right? 00:09:56.510 --> 00:09:59.700 align:middle line:90% 5, 7, 7, 7, 14, and 5? 00:09:59.700 --> 00:10:01.452 align:middle line:90% Yeah. 00:10:01.452 --> 00:10:02.160 align:middle line:90% I can't remember. 00:10:02.160 --> 00:10:04.890 align:middle line:84% It's been a long time since I wrote for 19 tone piano. 00:10:04.890 --> 00:10:08.090 align:middle line:84% They don't like tuning our pianos for that very much. 00:10:08.090 --> 00:10:12.800 align:middle line:84% But yeah, so it makes it a little bit more human there. 00:10:12.800 --> 00:10:14.695 align:middle line:90% Other choices you could do? 00:10:14.695 --> 00:10:20.390 align:middle line:90% 00:10:20.390 --> 00:10:22.100 align:middle line:90% Instruments. 00:10:22.100 --> 00:10:25.640 align:middle line:84% You could do how long it goes, whether it's 00:10:25.640 --> 00:10:27.390 align:middle line:90% going to be slower, faster. 00:10:27.390 --> 00:10:30.700 align:middle line:84% So these are some things we can do. 00:10:30.700 --> 00:10:36.210 align:middle line:84% So then we'll get into non-deterministic things. 00:10:36.210 --> 00:10:38.130 align:middle line:90% And so let's code one up first. 00:10:38.130 --> 00:10:41.930 align:middle line:84% So go and grab it before I get to particular pieces that 00:10:41.930 --> 00:10:42.625 align:middle line:90% are like that. 00:10:42.625 --> 00:10:47.090 align:middle line:90% 00:10:47.090 --> 00:10:49.580 align:middle line:84% So let's try some of the things that we might do 00:10:49.580 --> 00:10:52.920 align:middle line:84% and see how aesthetically interesting it is. 00:10:52.920 --> 00:10:53.870 align:middle line:90% Oops. 00:10:53.870 --> 00:10:55.880 align:middle line:90% Where'd that go? 00:10:55.880 --> 00:10:56.725 align:middle line:90% Is that over here? 00:10:56.725 --> 00:10:59.420 align:middle line:90% 00:10:59.420 --> 00:11:03.720 align:middle line:84% OK, gotta look, move this over as much as I can. 00:11:03.720 --> 00:11:05.480 align:middle line:90% OK, good. 00:11:05.480 --> 00:11:10.440 align:middle line:84% Great, so what's the module you hope I've unlocked? 00:11:10.440 --> 00:11:11.640 align:middle line:90% It's not part of Music21. 00:11:11.640 --> 00:11:13.640 align:middle line:84% It's part of Python, in order to do 00:11:13.640 --> 00:11:15.392 align:middle line:90% non-deterministic composition. 00:11:15.392 --> 00:11:16.100 align:middle line:90% AUDIENCE: Random. 00:11:16.100 --> 00:11:17.642 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: Random. 00:11:17.642 --> 00:11:19.880 align:middle line:90% Great, so let's import random. 00:11:19.880 --> 00:11:24.260 align:middle line:84% And for now, let's just import everything from Music21. 00:11:24.260 --> 00:11:25.152 align:middle line:90% Yep, that's good. 00:11:25.152 --> 00:11:25.860 align:middle line:90% You can see that. 00:11:25.860 --> 00:11:28.640 align:middle line:90% Good. 00:11:28.640 --> 00:11:33.170 align:middle line:84% Who has worked with the random module before? 00:11:33.170 --> 00:11:35.490 align:middle line:90% No? 00:11:35.490 --> 00:11:37.100 align:middle line:90% There's basically everybody. 00:11:37.100 --> 00:11:43.827 align:middle line:84% You know that random dot random gives you-- 00:11:43.827 --> 00:11:44.910 align:middle line:90% AUDIENCE: Between 0 and 1. 00:11:44.910 --> 00:11:46.650 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: Between 0 and 1, yep. 00:11:46.650 --> 00:11:47.530 align:middle line:90% Oh, shoot. 00:11:47.530 --> 00:11:49.050 align:middle line:90% No, that's not the same as-- 00:11:49.050 --> 00:11:50.910 align:middle line:90% I might have mistyped. 00:11:50.910 --> 00:11:55.722 align:middle line:84% I have it here, and it says it should be 0.23. 00:11:55.722 --> 00:11:56.430 align:middle line:90% Good, good, good. 00:11:56.430 --> 00:12:00.370 align:middle line:84% We all know that random-- so in a lot of languages, of course, 00:12:00.370 --> 00:12:04.030 align:middle line:84% you'll do, in order to get between 0 and 100, 00:12:04.030 --> 00:12:06.218 align:middle line:84% say, you would just multiply by 100. 00:12:06.218 --> 00:12:07.510 align:middle line:90% Oh, we did pretty well on that. 00:12:07.510 --> 00:12:09.780 align:middle line:90% Played the lottery today. 00:12:09.780 --> 00:12:11.410 align:middle line:90% You just multiply by 100. 00:12:11.410 --> 00:12:17.800 align:middle line:84% Python also has this randint from 1 to 100. 00:12:17.800 --> 00:12:22.720 align:middle line:84% And this is one of the older modules in Python. 00:12:22.720 --> 00:12:29.040 align:middle line:84% So it has a little bit of inconsistency 00:12:29.040 --> 00:12:35.070 align:middle line:84% with the rest of Python, in that, well, 00:12:35.070 --> 00:12:37.260 align:middle line:84% if I do this three times on average, 00:12:37.260 --> 00:12:42.330 align:middle line:84% I should be able to see the inconsistency that it includes 00:12:42.330 --> 00:12:43.390 align:middle line:90% the final number. 00:12:43.390 --> 00:12:46.870 align:middle line:84% So if you do range 1 to 100 in Python, 00:12:46.870 --> 00:12:49.660 align:middle line:84% your last one will be 99 in most languages. 00:12:49.660 --> 00:12:53.040 align:middle line:90% But this one will include that. 00:12:53.040 --> 00:12:57.870 align:middle line:84% Really useful-- this might be all review for somebody-- 00:12:57.870 --> 00:13:03.070 align:middle line:84% is that you can do random choice. 00:13:03.070 --> 00:13:06.670 align:middle line:84% So if you have a list of things or any other algorithm, 00:13:06.670 --> 00:13:10.420 align:middle line:84% you can randomly choose one from that. 00:13:10.420 --> 00:13:16.230 align:middle line:84% And you can do a weighed choice in this, 00:13:16.230 --> 00:13:21.690 align:middle line:90% simply by putting in something-- 00:13:21.690 --> 00:13:24.960 align:middle line:84% oops, I hit the wrong key-- putting in something 00:13:24.960 --> 00:13:26.045 align:middle line:90% a lot of times. 00:13:26.045 --> 00:13:29.310 align:middle line:90% 00:13:29.310 --> 00:13:32.200 align:middle line:84% Probably, if you're going to do this 10,000 by 10,000, 00:13:32.200 --> 00:13:35.140 align:middle line:84% there's other more efficient algorithms you can do for this. 00:13:35.140 --> 00:13:38.230 align:middle line:90% 00:13:38.230 --> 00:13:42.080 align:middle line:84% Questions on randomness outside of music? 00:13:42.080 --> 00:13:45.100 align:middle line:84% By the way, there's a way that you can set it 00:13:45.100 --> 00:13:48.010 align:middle line:84% up so that it's random at first, but then you 00:13:48.010 --> 00:13:51.040 align:middle line:84% can reconstruct your randomness by doing 00:13:51.040 --> 00:13:53.830 align:middle line:84% what's called a deterministic seed, which I'll often 00:13:53.830 --> 00:13:56.815 align:middle line:84% do when I'm writing compositions, 00:13:56.815 --> 00:13:58.940 align:middle line:84% so that if there's something I really, really like, 00:13:58.940 --> 00:14:02.070 align:middle line:84% I can reconstruct up until the point I didn't like it. 00:14:02.070 --> 00:14:05.090 align:middle line:90% 00:14:05.090 --> 00:14:08.290 align:middle line:84% So let's write, first off, what we call a random walk 00:14:08.290 --> 00:14:09.410 align:middle line:90% composition. 00:14:09.410 --> 00:14:12.850 align:middle line:84% So we're just going to start by doing something pretty stupid 00:14:12.850 --> 00:14:14.060 align:middle line:90% and try to make it better. 00:14:14.060 --> 00:14:17.500 align:middle line:84% So we're going to randomly walk around the scale. 00:14:17.500 --> 00:14:18.590 align:middle line:90% We'll start. 00:14:18.590 --> 00:14:22.360 align:middle line:84% Let's create something that's going 00:14:22.360 --> 00:14:26.380 align:middle line:84% to be a useful, helpful function, which 00:14:26.380 --> 00:14:27.470 align:middle line:90% I'll call it Multishow. 00:14:27.470 --> 00:14:30.110 align:middle line:84% You can skip the annotations while you're doing this. 00:14:30.110 --> 00:14:32.950 align:middle line:84% So what this does is just going to take in a stream 00:14:32.950 --> 00:14:39.930 align:middle line:84% and show it as both an image and MIDI. 00:14:39.930 --> 00:14:42.640 align:middle line:90% 00:14:42.640 --> 00:14:43.140 align:middle line:90% Bach. 00:14:43.140 --> 00:14:45.530 align:middle line:84% We'll just do our favorite Bach piece just to demo. 00:14:45.530 --> 00:14:50.540 align:middle line:90% BWV 66.6, multishow, Bach. 00:14:50.540 --> 00:14:55.630 align:middle line:84% This I find helpful for anytime I'm working with something. 00:14:55.630 --> 00:14:58.610 align:middle line:84% It always takes a lot longer when I'm projecting 00:14:58.610 --> 00:15:03.860 align:middle line:84% and the first time, but you get the score and then below. 00:15:03.860 --> 00:15:06.480 align:middle line:84% That never looks right on the video. 00:15:06.480 --> 00:15:08.486 align:middle line:90% Oops. 00:15:08.486 --> 00:15:11.930 align:middle line:90% [MUSIC PLAYING] 00:15:11.930 --> 00:15:15.471 align:middle line:90% Yep, that's the piece we know. 00:15:15.471 --> 00:15:17.580 align:middle line:84% So that's just a useful, helpful function. 00:15:17.580 --> 00:15:27.770 align:middle line:84% So one of the parts of pitches that I find really 00:15:27.770 --> 00:15:30.620 align:middle line:84% helpful for not being completely random 00:15:30.620 --> 00:15:33.260 align:middle line:84% here is this thing that would have 00:15:33.260 --> 00:15:37.220 align:middle line:84% been really helpful on that problem set where you had 00:15:37.220 --> 00:15:39.510 align:middle line:84% to determine the number of lines, 00:15:39.510 --> 00:15:43.550 align:middle line:84% the generic intervals between notes, if you remember that. 00:15:43.550 --> 00:15:46.520 align:middle line:84% That's something called diatonic note number, 00:15:46.520 --> 00:15:47.820 align:middle line:90% or diatonic note num. 00:15:47.820 --> 00:15:52.710 align:middle line:84% And this is a concept that I was really, really surprised. 00:15:52.710 --> 00:15:54.600 align:middle line:84% I was like, somebody must have named this, 00:15:54.600 --> 00:15:57.000 align:middle line:84% and it didn't seem to have a name before. 00:15:57.000 --> 00:16:01.842 align:middle line:84% It's simply, just like MIDI has-- 00:16:01.842 --> 00:16:03.800 align:middle line:84% I don't think I went over this in class, right? 00:16:03.800 --> 00:16:07.010 align:middle line:84% No, I don't think so-- that C4, middle C, 00:16:07.010 --> 00:16:12.920 align:middle line:84% is arbitrarily assigned to MIDI number 60, 00:16:12.920 --> 00:16:18.090 align:middle line:84% arbitrarily assigned middle C to diatonic note number-- 00:16:18.090 --> 00:16:20.138 align:middle line:90% that's MIDI 60-- 00:16:20.138 --> 00:16:21.720 align:middle line:90% 29. 00:16:21.720 --> 00:16:24.030 align:middle line:84% Neither of these are completely arbitrary. 00:16:24.030 --> 00:16:25.260 align:middle line:90% There's a reason for them. 00:16:25.260 --> 00:16:27.510 align:middle line:90% And the notion is that all. 00:16:27.510 --> 00:16:32.170 align:middle line:84% Cs, C sharps, et cetera, have diatonic note num 29. 00:16:32.170 --> 00:16:37.410 align:middle line:84% So C sharp, and that just tells you-- 00:16:37.410 --> 00:16:42.930 align:middle line:84% C double sharp, yep, so on, C flat. 00:16:42.930 --> 00:16:44.070 align:middle line:90% Great. 00:16:44.070 --> 00:16:47.310 align:middle line:84% And so this is helpful for writing diatonic music, 00:16:47.310 --> 00:16:51.100 align:middle line:84% that you can just manipulate the diatonic note number of a note 00:16:51.100 --> 00:16:56.257 align:middle line:84% and you don't have to worry making it sound a little bit 00:16:56.257 --> 00:16:57.090 align:middle line:90% better or something. 00:16:57.090 --> 00:16:59.160 align:middle line:84% We should have a pentatonic note number, 00:16:59.160 --> 00:17:03.430 align:middle line:84% also, because that makes it even better to do something. 00:17:03.430 --> 00:17:05.310 align:middle line:90% So what I'm going to do is-- 00:17:05.310 --> 00:17:11.910 align:middle line:84% and then we can just do n dot pitch, dot diatonic note num, 00:17:11.910 --> 00:17:20.520 align:middle line:84% plus equals 4, and move things around that way. 00:17:20.520 --> 00:17:25.260 align:middle line:84% If you haven't seen it yet, one of the things in Music21 that's 00:17:25.260 --> 00:17:29.880 align:middle line:84% not so good is that you'll often need to copy things, 00:17:29.880 --> 00:17:31.630 align:middle line:84% so we'll get to that in just a little bit. 00:17:31.630 --> 00:17:33.640 align:middle line:90% But I'm going to import, copy. 00:17:33.640 --> 00:17:39.580 align:middle line:84% And if you've never used this copy n-- 00:17:39.580 --> 00:17:40.080 align:middle line:90% oops. 00:17:40.080 --> 00:17:41.040 align:middle line:90% Copy is a module. 00:17:41.040 --> 00:17:42.060 align:middle line:90% Copy dot copy. 00:17:42.060 --> 00:17:46.350 align:middle line:90% 00:17:46.350 --> 00:17:49.570 align:middle line:84% So I did turn n into a g, and so now it's a g. 00:17:49.570 --> 00:17:51.270 align:middle line:90% There's two things. 00:17:51.270 --> 00:17:55.830 align:middle line:84% Who has used the copy module in Python before? 00:17:55.830 --> 00:17:56.620 align:middle line:90% No, not too many. 00:17:56.620 --> 00:17:57.520 align:middle line:90% OK, good. 00:17:57.520 --> 00:18:00.450 align:middle line:84% So because for the most part, you 00:18:00.450 --> 00:18:04.058 align:middle line:84% might just want to just create another note 00:18:04.058 --> 00:18:05.100 align:middle line:90% if you wanted to do that. 00:18:05.100 --> 00:18:06.540 align:middle line:90% Don't need to type that. 00:18:06.540 --> 00:18:09.430 align:middle line:84% But you can also copy an existing one. 00:18:09.430 --> 00:18:15.420 align:middle line:84% And what you'll use much more is something called deep copy. 00:18:15.420 --> 00:18:18.180 align:middle line:84% And the difference, you can see that copying the note 00:18:18.180 --> 00:18:20.410 align:middle line:90% gives you a note that's g. 00:18:20.410 --> 00:18:22.660 align:middle line:84% Deep copying gives it a note that's g. 00:18:22.660 --> 00:18:27.450 align:middle line:84% The difference is that, well, here. 00:18:27.450 --> 00:18:30.730 align:middle line:90% Easier to show than tell. 00:18:30.730 --> 00:18:40.390 align:middle line:84% And c is going to be the copy of n, and we'll say dc is the d-- 00:18:40.390 --> 00:18:42.290 align:middle line:90% my computer just froze. 00:18:42.290 --> 00:18:43.330 align:middle line:90% There we go. 00:18:43.330 --> 00:18:45.360 align:middle line:90% dc is the deep copy of n. 00:18:45.360 --> 00:18:49.030 align:middle line:90% 00:18:49.030 --> 00:18:51.670 align:middle line:84% The difference is, so here, we'll just 00:18:51.670 --> 00:18:56.200 align:middle line:84% make sure that they're all here, they're all, yep, three notes. 00:18:56.200 --> 00:19:03.970 align:middle line:84% Difference is if we take n's pitch diatonic note num 00:19:03.970 --> 00:19:17.260 align:middle line:84% and go up 4 more, that the copy of n also changes, 00:19:17.260 --> 00:19:21.980 align:middle line:84% whereas the deep copy doesn't change. 00:19:21.980 --> 00:19:27.790 align:middle line:84% And so what some people might think of this as, you end up 00:19:27.790 --> 00:19:30.790 align:middle line:84% working with arrays in a lot of other languages 00:19:30.790 --> 00:19:32.800 align:middle line:84% and stuff that copying an array, you still 00:19:32.800 --> 00:19:34.970 align:middle line:84% have references to the internal pointers. 00:19:34.970 --> 00:19:38.350 align:middle line:84% So anything that is itself an object, 00:19:38.350 --> 00:19:46.660 align:middle line:84% like n.pitch and nc.pitch will still be the same 00:19:46.660 --> 00:19:48.495 align:middle line:90% under a shallow or normal copy. 00:19:48.495 --> 00:19:54.520 align:middle line:90% 00:19:54.520 --> 00:19:59.080 align:middle line:84% But with a deep copy, they're different. 00:19:59.080 --> 00:20:01.960 align:middle line:90% Deep copies, pitch. 00:20:01.960 --> 00:20:05.200 align:middle line:84% So you'll find when you're manipulating things 00:20:05.200 --> 00:20:07.750 align:middle line:84% and you get to a particular point in the score, 00:20:07.750 --> 00:20:11.410 align:middle line:84% and now you want to add something else that's 00:20:11.410 --> 00:20:13.840 align:middle line:84% a manipulation of what came before, you'll want to make 00:20:13.840 --> 00:20:16.180 align:middle line:90% a deep copy of it beforehand. 00:20:16.180 --> 00:20:17.090 align:middle line:90% Does that make sense? 00:20:17.090 --> 00:20:18.490 align:middle line:90% Cool? 00:20:18.490 --> 00:20:19.100 align:middle line:90% Any questions? 00:20:19.100 --> 00:20:19.390 align:middle line:90% Yeah. 00:20:19.390 --> 00:20:21.265 align:middle line:84% AUDIENCE: What did you show in the last part? 00:20:21.265 --> 00:20:24.585 align:middle line:84% It's like the pitch is going to be different by definition? 00:20:24.585 --> 00:20:26.460 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: Oh, sure, sure. 00:20:26.460 --> 00:20:29.580 align:middle line:84% I think it's a little bit hard because this is this part here. 00:20:29.580 --> 00:20:32.000 align:middle line:84% So when you make a shallow copy and you 00:20:32.000 --> 00:20:36.870 align:middle line:84% change some nested structure within there, 00:20:36.870 --> 00:20:41.690 align:middle line:84% like a pitch or a duration, it will not be-- 00:20:41.690 --> 00:20:43.320 align:middle line:90% they're all shared. 00:20:43.320 --> 00:20:47.898 align:middle line:84% So what you do is think of it as-- 00:20:47.898 --> 00:20:48.440 align:middle line:90% I don't know. 00:20:48.440 --> 00:20:50.390 align:middle line:84% What's something else that's not so? 00:20:50.390 --> 00:20:53.480 align:middle line:90% So n is a note and n has-- 00:20:53.480 --> 00:20:55.070 align:middle line:90% I wish I had set this before. 00:20:55.070 --> 00:21:03.020 align:middle line:84% ID equals note, and dot pitch equals some of a pitch dot pitch 00:21:03.020 --> 00:21:03.520 align:middle line:90% object. 00:21:03.520 --> 00:21:07.160 align:middle line:90% 00:21:07.160 --> 00:21:15.320 align:middle line:84% And dot duration, let's say, dur is some kind of duration object. 00:21:15.320 --> 00:21:21.930 align:middle line:84% When you do a shallow copy and c will be nc's ID 00:21:21.930 --> 00:21:32.000 align:middle line:84% equals n's ID, and nc's pitch equals nc, n's pitch. 00:21:32.000 --> 00:21:34.530 align:middle line:90% 00:21:34.530 --> 00:21:39.550 align:middle line:84% And so since ID is just a string or something like that, 00:21:39.550 --> 00:21:44.280 align:middle line:84% once it's copied over, they have no connection to each other. 00:21:44.280 --> 00:21:47.610 align:middle line:84% But since we've copied the pitch over, 00:21:47.610 --> 00:21:51.870 align:middle line:84% and it's a complex object when you change something 00:21:51.870 --> 00:21:56.220 align:middle line:84% on this, when this manipulates, this manipulates. 00:21:56.220 --> 00:22:03.330 align:middle line:84% So with a deep copy, what you do is dc dot pitch. 00:22:03.330 --> 00:22:04.180 align:middle line:90% Hold that. 00:22:04.180 --> 00:22:06.970 align:middle line:84% First let's do a deep copy of the pitch. 00:22:06.970 --> 00:22:08.790 align:middle line:84% And so you keep going down and you 00:22:08.790 --> 00:22:11.765 align:middle line:84% make sure that nothing is connected between the two. 00:22:11.765 --> 00:22:14.660 align:middle line:90% 00:22:14.660 --> 00:22:17.020 align:middle line:84% AUDIENCE: So this also is slightly misleading, right? 00:22:17.020 --> 00:22:20.190 align:middle line:84% Because there's nothing that actually says nc.pitch pitch 00:22:20.190 --> 00:22:22.360 align:middle line:90% is equal to n.pitch. 00:22:22.360 --> 00:22:24.720 align:middle line:90% In fact, that's the whole point. 00:22:24.720 --> 00:22:27.070 align:middle line:84% If it did, that would be an example of deep copy, right? 00:22:27.070 --> 00:22:28.445 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: No, 00:22:28.445 --> 00:22:30.970 align:middle line:84% because you're referencing the same object in point. 00:22:30.970 --> 00:22:33.750 align:middle line:84% So when you say n dot pitch, it's actually 00:22:33.750 --> 00:22:35.910 align:middle line:84% thinking in the computer, oh, that 00:22:35.910 --> 00:22:41.430 align:middle line:84% is the complex object being stored at some random memory 00:22:41.430 --> 00:22:42.437 align:middle line:90% location somewhere. 00:22:42.437 --> 00:22:44.520 align:middle line:84% AUDIENCE: OK, so it's still pointing to the label. 00:22:44.520 --> 00:22:44.820 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: So it's still 00:22:44.820 --> 00:22:46.030 align:middle line:90% pointing to the same label. 00:22:46.030 --> 00:22:47.380 align:middle line:90% Yeah, so exactly. 00:22:47.380 --> 00:22:51.840 align:middle line:84% For people who know that kind of terminology, 00:22:51.840 --> 00:22:54.150 align:middle line:84% it's pointing to the same label, whereas this one 00:22:54.150 --> 00:22:56.130 align:middle line:84% goes to the label, and copies everything 00:22:56.130 --> 00:23:00.120 align:middle line:90% and makes a new pointer. 00:23:00.120 --> 00:23:03.160 align:middle line:84% So that ends up being pretty important. 00:23:03.160 --> 00:23:07.710 align:middle line:84% And if anybody wants to make their final project making 00:23:07.710 --> 00:23:10.350 align:middle line:84% Music21 faster, you could find places 00:23:10.350 --> 00:23:12.990 align:middle line:84% that we deep copy more than we have to 00:23:12.990 --> 00:23:18.280 align:middle line:84% because that's the slowest thing in what's happening. 00:23:18.280 --> 00:23:19.820 align:middle line:90% Too much of an aside. 00:23:19.820 --> 00:23:21.380 align:middle line:90% Back to some music. 00:23:21.380 --> 00:23:21.940 align:middle line:90% Great. 00:23:21.940 --> 00:23:25.220 align:middle line:84% So we'll get back to our random walk composition 00:23:25.220 --> 00:23:28.660 align:middle line:90% once I find where this went. 00:23:28.660 --> 00:23:30.850 align:middle line:90% Yeah, here we go. 00:23:30.850 --> 00:23:34.340 align:middle line:84% Great, so we're going to be copying things pretty soon. 00:23:34.340 --> 00:23:36.680 align:middle line:90% But first, let's just create. 00:23:36.680 --> 00:23:40.840 align:middle line:84% We'll create a generic stream, which we rarely do nowadays. 00:23:40.840 --> 00:23:42.830 align:middle line:84% Maybe we want to make it a part or something. 00:23:42.830 --> 00:23:44.930 align:middle line:84% But generic stream will work pretty well. 00:23:44.930 --> 00:23:50.740 align:middle line:84% And let's just say for i in range 30, let's just append 00:23:50.740 --> 00:23:53.230 align:middle line:90% 30 random notes. 00:23:53.230 --> 00:23:56.720 align:middle line:90% Or no, let's do this. 00:23:56.720 --> 00:24:00.130 align:middle line:90% No, let's start on c, current. 00:24:00.130 --> 00:24:01.640 align:middle line:90% Well, we'll just call it cur. 00:24:01.640 --> 00:24:04.330 align:middle line:84% That's the current diatonic note number. 00:24:04.330 --> 00:24:07.300 align:middle line:84% When we're coding fast, we don't need my obsession 00:24:07.300 --> 00:24:08.180 align:middle line:90% with long variables. 00:24:08.180 --> 00:24:13.750 align:middle line:84% So we'll say note equals note dot note, and pitch 00:24:13.750 --> 00:24:19.300 align:middle line:84% diatonic note num equals the current one. 00:24:19.300 --> 00:24:21.280 align:middle line:90% And that works pretty well. 00:24:21.280 --> 00:24:22.795 align:middle line:90% Stream, append. 00:24:22.795 --> 00:24:24.500 align:middle line:90% AUDIENCE: It's called cur now. 00:24:24.500 --> 00:24:25.990 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: Sorry, thank you. 00:24:25.990 --> 00:24:27.282 align:middle line:90% That's why I shouldn't do that. 00:24:27.282 --> 00:24:27.890 align:middle line:90% Thank you. 00:24:27.890 --> 00:24:29.630 align:middle line:90% Please, always just do that. 00:24:29.630 --> 00:24:30.460 align:middle line:90% I appreciate that. 00:24:30.460 --> 00:24:31.793 align:middle line:90% Big help. 00:24:31.793 --> 00:24:33.460 align:middle line:84% And we'll just say we're going to choose 00:24:33.460 --> 00:24:37.130 align:middle line:84% between going down a step, going up a step, and so on. 00:24:37.130 --> 00:24:39.800 align:middle line:84% So our first note will always be-- 00:24:39.800 --> 00:24:40.300 align:middle line:90% AUDIENCE: C. 00:24:40.300 --> 00:24:41.960 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: C, C4. 00:24:41.960 --> 00:24:43.890 align:middle line:90% And now let's multi-show that. 00:24:43.890 --> 00:24:46.570 align:middle line:90% 00:24:46.570 --> 00:24:50.810 align:middle line:84% Pretty boring piece, but here we go. 00:24:50.810 --> 00:24:53.120 align:middle line:84% Oh, this one went down this time. 00:24:53.120 --> 00:24:56.431 align:middle line:90% [MUSICAL NOTES] 00:24:56.431 --> 00:24:58.330 align:middle line:90% 00:24:58.330 --> 00:24:58.840 align:middle line:90% Et cetera. 00:24:58.840 --> 00:25:00.530 align:middle line:90% Anybody's more interesting? 00:25:00.530 --> 00:25:02.290 align:middle line:90% Anyone's go up? 00:25:02.290 --> 00:25:05.140 align:middle line:90% Anyone's just hit Play? 00:25:05.140 --> 00:25:06.280 align:middle line:90% Yeah. 00:25:06.280 --> 00:25:08.890 align:middle line:90% So not a great piece. 00:25:08.890 --> 00:25:10.450 align:middle line:84% The easiest thing I could do to make 00:25:10.450 --> 00:25:12.350 align:middle line:84% it a little bit better, by the way, 00:25:12.350 --> 00:25:15.490 align:middle line:84% would be to make my number of notes I attach be-- 00:25:15.490 --> 00:25:16.530 align:middle line:90% what do you call it? 00:25:16.530 --> 00:25:19.071 align:middle line:90% A multiple of 4. 00:25:19.071 --> 00:25:19.830 align:middle line:90% But OK. 00:25:19.830 --> 00:25:21.860 align:middle line:84% So let's move on to do some things that are 00:25:21.860 --> 00:25:24.590 align:middle line:90% a little bit more interesting. 00:25:24.590 --> 00:25:28.490 align:middle line:84% Maybe we could do something like this. 00:25:28.490 --> 00:25:31.170 align:middle line:84% I'm going to copy and paste a lot on this, 00:25:31.170 --> 00:25:35.660 align:middle line:84% so we can go ahead and just tweak what we're doing. 00:25:35.660 --> 00:25:38.430 align:middle line:84% You, in fact, probably don't even need to copy and paste. 00:25:38.430 --> 00:25:40.530 align:middle line:84% You can just keep tweaking the same cell forever. 00:25:40.530 --> 00:25:43.190 align:middle line:84% So what's something we should do to make this a little bit more 00:25:43.190 --> 00:25:44.130 align:middle line:90% interesting? 00:25:44.130 --> 00:25:49.310 align:middle line:90% 00:25:49.310 --> 00:25:50.660 align:middle line:90% Rhythm. 00:25:50.660 --> 00:25:51.390 align:middle line:90% Great. 00:25:51.390 --> 00:25:57.068 align:middle line:84% So let's have each note and dot duration, dot something. 00:25:57.068 --> 00:25:57.985 align:middle line:90% What do we want to do? 00:25:57.985 --> 00:26:04.700 align:middle line:90% 00:26:04.700 --> 00:26:07.280 align:middle line:84% AUDIENCE: I'm just thinking about something for-- 00:26:07.280 --> 00:26:08.490 align:middle line:90% it wasn't for a duration. 00:26:08.490 --> 00:26:09.170 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: OK, then hold 00:26:09.170 --> 00:26:12.390 align:middle line:84% it when we get-- let's finish the duration, and then yeah, 00:26:12.390 --> 00:26:14.240 align:middle line:90% then we'll get to it. 00:26:14.240 --> 00:26:14.870 align:middle line:90% John? 00:26:14.870 --> 00:26:20.000 align:middle line:84% AUDIENCE: A choice for 4s to choose a quarter, a half, 00:26:20.000 --> 00:26:21.780 align:middle line:90% eighth, n for the duration. 00:26:21.780 --> 00:26:23.280 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: Great. 00:26:23.280 --> 00:26:25.100 align:middle line:90% So random choice. 00:26:25.100 --> 00:26:26.810 align:middle line:84% While I start typing this out, what 00:26:26.810 --> 00:26:34.760 align:middle line:84% goes after the n dot duration dot quarter, half, eighth? 00:26:34.760 --> 00:26:39.080 align:middle line:84% Anyone remember what this is called? 00:26:39.080 --> 00:26:39.750 align:middle line:90% It's unhelpful. 00:26:39.750 --> 00:26:41.460 align:middle line:90% It's called the type. 00:26:41.460 --> 00:26:43.010 align:middle line:84% The type quarter length, we would 00:26:43.010 --> 00:26:48.030 align:middle line:84% have to say we could put it in as one, two, half. 00:26:48.030 --> 00:26:50.300 align:middle line:90% Yep. 00:26:50.300 --> 00:26:55.296 align:middle line:84% Let's see if that makes it a little bit nicer. 00:26:55.296 --> 00:26:58.790 align:middle line:90% [MUSICAL NOTES] 00:26:58.790 --> 00:27:01.520 align:middle line:90% It's actually already nicer. 00:27:01.520 --> 00:27:03.390 align:middle line:84% Usually, this doesn't help as much. 00:27:03.390 --> 00:27:06.950 align:middle line:90% [MUSICAL NOTES] 00:27:06.950 --> 00:27:08.720 align:middle line:90% 00:27:08.720 --> 00:27:11.610 align:middle line:84% Next improvement I'm just going to keep going on this. 00:27:11.610 --> 00:27:12.420 align:middle line:90% Yeah. 00:27:12.420 --> 00:27:14.850 align:middle line:84% AUDIENCE: You can make it [INAUDIBLE]. 00:27:14.850 --> 00:27:15.630 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: What's that? 00:27:15.630 --> 00:27:16.950 align:middle line:84% AUDIENCE: You can make it go up a third. 00:27:16.950 --> 00:27:17.615 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: Cool. 00:27:17.615 --> 00:27:19.780 align:middle line:84% Yep, it can go up a third or down a third. 00:27:19.780 --> 00:27:23.975 align:middle line:84% Now, do we want to make it a weighed probability at all? 00:27:23.975 --> 00:27:27.210 align:middle line:90% 00:27:27.210 --> 00:27:28.710 align:middle line:90% So how much more do we want it? 00:27:28.710 --> 00:27:32.740 align:middle line:84% Up a third is minus 2, because we're thinking in computer 00:27:32.740 --> 00:27:34.830 align:middle line:84% think, not in music theory think. 00:27:34.830 --> 00:27:36.506 align:middle line:90% So that would be unweighed. 00:27:36.506 --> 00:27:40.170 align:middle line:90% 00:27:40.170 --> 00:27:46.620 align:middle line:84% And how do we want to weigh this to make it feel more musical? 00:27:46.620 --> 00:27:48.460 align:middle line:84% AUDIENCE: Higher weight on stepwise motion? 00:27:48.460 --> 00:27:50.793 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: Higher weight on stepwise. 00:27:50.793 --> 00:27:52.320 align:middle line:90% 00:27:52.320 --> 00:27:54.930 align:middle line:90% Negative 1, negative 1, 1, 1. 00:27:54.930 --> 00:27:57.005 align:middle line:90% OK, let's try that. 00:27:57.005 --> 00:28:01.912 align:middle line:90% 00:28:01.912 --> 00:28:05.398 align:middle line:90% [MUSICAL NOTES] 00:28:05.398 --> 00:28:12.370 align:middle line:90% 00:28:12.370 --> 00:28:13.950 align:middle line:90% Hey, there's our first step. 00:28:13.950 --> 00:28:16.790 align:middle line:90% [MUSICAL NOTES] 00:28:16.790 --> 00:28:17.690 align:middle line:90% Not so good. 00:28:17.690 --> 00:28:19.390 align:middle line:84% OK, so we still need some things. 00:28:19.390 --> 00:28:20.660 align:middle line:90% That's definite improvement. 00:28:20.660 --> 00:28:21.790 align:middle line:90% Thank you. 00:28:21.790 --> 00:28:23.640 align:middle line:90% Another thing to improve. 00:28:23.640 --> 00:28:27.550 align:middle line:90% 00:28:27.550 --> 00:28:28.223 align:middle line:90% Yeah, Jason. 00:28:28.223 --> 00:28:30.640 align:middle line:84% AUDIENCE: Decrease the weight on staying on the same note? 00:28:30.640 --> 00:28:31.570 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: Decrease the weight 00:28:31.570 --> 00:28:32.737 align:middle line:90% on staying on the same note. 00:28:32.737 --> 00:28:35.680 align:middle line:84% OK, so what we can do is the easiest way 00:28:35.680 --> 00:28:38.090 align:middle line:84% when you're programming really fast in front of a class-- 00:28:38.090 --> 00:28:40.560 align:middle line:90% do that-- is cut and paste. 00:28:40.560 --> 00:28:42.560 align:middle line:84% Don't do this in your actual things you turn in. 00:28:42.560 --> 00:28:46.070 align:middle line:84% But cut and paste and just delete that 0. 00:28:46.070 --> 00:28:46.650 align:middle line:90% Is that good? 00:28:46.650 --> 00:28:52.750 align:middle line:90% 00:28:52.750 --> 00:28:54.890 align:middle line:84% Yeah, it already looks a little bit better. 00:28:54.890 --> 00:28:58.690 align:middle line:84% Next, let's do another improvement on rhythm. 00:28:58.690 --> 00:28:59.740 align:middle line:90% Yeah. 00:28:59.740 --> 00:29:02.050 align:middle line:84% AUDIENCE: It's more likely you have 00:29:02.050 --> 00:29:05.912 align:middle line:90% a note starting on beat one. 00:29:05.912 --> 00:29:07.370 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: Good. 00:29:07.370 --> 00:29:09.850 align:middle line:84% So we want to try to weigh it somehow 00:29:09.850 --> 00:29:14.440 align:middle line:84% so that we have notes that start on the downbeat more. 00:29:14.440 --> 00:29:17.810 align:middle line:84% Talk with your neighbor on how you might implement that. 00:29:17.810 --> 00:29:19.072 align:middle line:90% [INTERPOSING VOICES] 00:29:19.072 --> 00:29:21.530 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: OK, let's come back together. 00:29:21.530 --> 00:29:22.202 align:middle line:90% Anybody? 00:29:22.202 --> 00:29:24.085 align:middle line:90% [INTERPOSING VOICES] 00:29:24.085 --> 00:29:25.460 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: OK. 00:29:25.460 --> 00:29:29.598 align:middle line:84% Let's jump in, well, and we'll try to program it together 00:29:29.598 --> 00:29:30.890 align:middle line:90% because we'll do this in class. 00:29:30.890 --> 00:29:32.680 align:middle line:84% Somebody I haven't heard from yet 00:29:32.680 --> 00:29:36.000 align:middle line:84% want to give a suggestion who hasn't unspoken? 00:29:36.000 --> 00:29:38.770 align:middle line:90% 00:29:38.770 --> 00:29:39.453 align:middle line:90% Yeah? 00:29:39.453 --> 00:29:40.870 align:middle line:84% AUDIENCE: We can try to keep track 00:29:40.870 --> 00:29:43.910 align:middle line:84% of an offset, and then based on that offset, have a different-- 00:29:43.910 --> 00:29:45.310 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: So have a different probability 00:29:45.310 --> 00:29:46.250 align:middle line:90% based on the offset. 00:29:46.250 --> 00:29:51.320 align:middle line:84% What offsets in currently with the default time signature 4/4? 00:29:51.320 --> 00:29:53.360 align:middle line:84% Why do we make that default very hegemonic. 00:29:53.360 --> 00:30:00.743 align:middle line:84% But what offsets represent the downbeat? 00:30:00.743 --> 00:30:07.250 align:middle line:90% 00:30:07.250 --> 00:30:08.060 align:middle line:90% AUDIENCE: Zero. 00:30:08.060 --> 00:30:09.080 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: Zero definitely 00:30:09.080 --> 00:30:10.080 align:middle line:90% represents a downbeat. 00:30:10.080 --> 00:30:11.840 align:middle line:90% What other offsets? 00:30:11.840 --> 00:30:14.550 align:middle line:84% So have we been creating measures, by the way. 00:30:14.550 --> 00:30:16.140 align:middle line:90% First off, let's look at that. 00:30:16.140 --> 00:30:20.540 align:middle line:84% So we've kind of taking advantage 00:30:20.540 --> 00:30:24.000 align:middle line:84% of Music21 or something creating measures for us. 00:30:24.000 --> 00:30:26.390 align:middle line:84% So our offsets, we start with zero, 00:30:26.390 --> 00:30:27.900 align:middle line:90% and then we're just appending. 00:30:27.900 --> 00:30:30.113 align:middle line:90% So what's our next downbeat? 00:30:30.113 --> 00:30:33.260 align:middle line:90% 00:30:33.260 --> 00:30:34.122 align:middle line:90% AUDIENCE: Four. 00:30:34.122 --> 00:30:35.580 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: Good. 00:30:35.580 --> 00:30:38.780 align:middle line:84% And somebody else, what's our general formula we're 00:30:38.780 --> 00:30:43.700 align:middle line:90% going to find for downbeat? 00:30:43.700 --> 00:30:46.130 align:middle line:90% AUDIENCE: 0 mod 4? 00:30:46.130 --> 00:30:48.950 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: Yeah, we 00:30:48.950 --> 00:30:55.402 align:middle line:84% could check if the offset is mod 4 is 0. 00:30:55.402 --> 00:31:00.560 align:middle line:84% So we'll get to, before we do that, so let's 00:31:00.560 --> 00:31:04.560 align:middle line:90% say, what's our current offset. 00:31:04.560 --> 00:31:05.990 align:middle line:84% So one of the things we can do is 00:31:05.990 --> 00:31:09.390 align:middle line:84% we can keep track of our current offset. 00:31:09.390 --> 00:31:11.570 align:middle line:84% Now, we're going to wish we had-- 00:31:11.570 --> 00:31:12.720 align:middle line:90% 0.0. 00:31:12.720 --> 00:31:15.690 align:middle line:84% Now we're going to wish we had done this as quarter lengths. 00:31:15.690 --> 00:31:21.600 align:middle line:84% So I'm going to rewrite this as quarter length. 00:31:21.600 --> 00:31:22.100 align:middle line:90% Why? 00:31:22.100 --> 00:31:24.260 align:middle line:84% Because it's easier to add quarter lengths than it 00:31:24.260 --> 00:31:24.930 align:middle line:90% is types. 00:31:24.930 --> 00:31:29.930 align:middle line:90% 00:31:29.930 --> 00:31:32.300 align:middle line:90% So 1, 2, 0.5. 00:31:32.300 --> 00:31:40.056 align:middle line:84% And we'll just say ql equals random.choice. 00:31:40.056 --> 00:31:43.430 align:middle line:90% 00:31:43.430 --> 00:31:48.105 align:middle line:84% So what we can do is, well, now, I'm 00:31:48.105 --> 00:31:50.480 align:middle line:84% just going to make sure that I still have a current thing 00:31:50.480 --> 00:31:57.530 align:middle line:90% and cur_offset plus equals ql. 00:31:57.530 --> 00:32:01.680 align:middle line:84% So I haven't done anything to adjust our probability yet. 00:32:01.680 --> 00:32:03.320 align:middle line:84% Just making sure that this works. 00:32:03.320 --> 00:32:04.570 align:middle line:90% Yep, it it's still generating. 00:32:04.570 --> 00:32:08.310 align:middle line:84% So how can we use this to implement the great suggestion 00:32:08.310 --> 00:32:12.540 align:middle line:84% that we want to have things that fit on the downbeat 00:32:12.540 --> 00:32:15.040 align:middle line:90% more often do something else? 00:32:15.040 --> 00:32:26.160 align:middle line:84% So we'll say, is on downbeat equals cur_offset present zero. 00:32:26.160 --> 00:32:29.850 align:middle line:90% You got to say, bools on. 00:32:29.850 --> 00:32:31.150 align:middle line:90% Here we say true. 00:32:31.150 --> 00:32:32.500 align:middle line:90% This is the easiest way. 00:32:32.500 --> 00:32:38.160 align:middle line:84% If cur offset 0 equals 0, else false. 00:32:38.160 --> 00:32:39.280 align:middle line:90% You can do it. 00:32:39.280 --> 00:32:41.340 align:middle line:84% You can realize that Boolean expression will 00:32:41.340 --> 00:32:46.650 align:middle line:84% be reduce itself, but I like to be explicit when I'm thinking. 00:32:46.650 --> 00:32:50.190 align:middle line:84% So now we know if something is on the downbeat. 00:32:50.190 --> 00:32:51.986 align:middle line:84% How do we want to do it differently? 00:32:51.986 --> 00:32:56.070 align:middle line:90% 00:32:56.070 --> 00:32:57.920 align:middle line:90% Do you need to brainstorm? 00:32:57.920 --> 00:33:02.380 align:middle line:90% 00:33:02.380 --> 00:33:03.080 align:middle line:90% Thanks, Adam. 00:33:03.080 --> 00:33:06.640 align:middle line:84% I want to hear from some more voices, 00:33:06.640 --> 00:33:10.210 align:middle line:90% even though you're doing great. 00:33:10.210 --> 00:33:10.953 align:middle line:90% Vanessa, yeah. 00:33:10.953 --> 00:33:12.370 align:middle line:84% AUDIENCE: On the downbeat, you can 00:33:12.370 --> 00:33:15.997 align:middle line:84% use the formally chosen borderline pitch, and then-- 00:33:15.997 --> 00:33:18.580 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: And if it's not on the downbeat? 00:33:18.580 --> 00:33:19.880 align:middle line:84% AUDIENCE: You leave it as what it was before. 00:33:19.880 --> 00:33:20.440 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: You leave it 00:33:20.440 --> 00:33:21.830 align:middle line:90% as what it was before. 00:33:21.830 --> 00:33:24.550 align:middle line:90% Let's try that. 00:33:24.550 --> 00:33:25.430 align:middle line:90% Oh, yeah. 00:33:25.430 --> 00:33:30.640 align:middle line:84% So we have a last ql, so we'll define ql outside of here, 00:33:30.640 --> 00:33:35.230 align:middle line:84% will be pretty Pythonic, equals-- 00:33:35.230 --> 00:33:37.460 align:middle line:90% we'll say our last one is 1. 00:33:37.460 --> 00:33:39.550 align:middle line:90% We're actually not going to. 00:33:39.550 --> 00:33:42.020 align:middle line:84% Because we will always start on the downbeat, 00:33:42.020 --> 00:33:43.313 align:middle line:90% ql will be defined. 00:33:43.313 --> 00:33:44.980 align:middle line:84% But let's do good programming techniques 00:33:44.980 --> 00:33:46.190 align:middle line:90% and define it outside. 00:33:46.190 --> 00:33:46.690 align:middle line:90% Good. 00:33:46.690 --> 00:33:52.373 align:middle line:84% So if is on, if not is on downbeat. 00:33:52.373 --> 00:33:57.730 align:middle line:90% 00:33:57.730 --> 00:33:59.110 align:middle line:90% Let's see if that-- 00:33:59.110 --> 00:34:01.300 align:middle line:84% otherwise, we're just going to stay 00:34:01.300 --> 00:34:04.000 align:middle line:90% on whatever we were on before. 00:34:04.000 --> 00:34:04.790 align:middle line:90% Let's see. 00:34:04.790 --> 00:34:06.737 align:middle line:90% Think it through first. 00:34:06.737 --> 00:34:07.820 align:middle line:90% Did I get something wrong? 00:34:07.820 --> 00:34:09.810 align:middle line:84% AUDIENCE: ql will have a not defined error. 00:34:09.810 --> 00:34:13.270 align:middle line:90% 00:34:13.270 --> 00:34:14.000 align:middle line:90% Oh, sorry. 00:34:14.000 --> 00:34:14.960 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: Yep, that's what we do. 00:34:14.960 --> 00:34:16.752 align:middle line:84% That's why I was talking about defining it. 00:34:16.752 --> 00:34:18.040 align:middle line:90% Good catch, though. 00:34:18.040 --> 00:34:22.969 align:middle line:84% Always don't know crashed 787 because of not defined errors. 00:34:22.969 --> 00:34:24.940 align:middle line:90% That's a famous story. 00:34:24.940 --> 00:34:28.780 align:middle line:84% OK, so let's make sure that we're not 00:34:28.780 --> 00:34:32.000 align:middle line:84% going to end up permanently syncopated or something. 00:34:32.000 --> 00:34:36.920 align:middle line:84% We start on the downbeat, and then we choose between-- 00:34:36.920 --> 00:34:43.219 align:middle line:84% maybe I should put these in some of a logical order 2, 1, 0.5. 00:34:43.219 --> 00:34:46.130 align:middle line:84% We randomly choose between half note, quarter note, 00:34:46.130 --> 00:34:48.469 align:middle line:84% and eighth note, and we're on the downbeat, 00:34:48.469 --> 00:34:51.500 align:middle line:84% and so we're going to stay on that for a bit. 00:34:51.500 --> 00:34:55.179 align:middle line:84% We'll always end up back on a downbeat at some point. 00:34:55.179 --> 00:34:56.270 align:middle line:90% We start on a downbeat. 00:34:56.270 --> 00:35:03.200 align:middle line:84% We choose 2, 1, 0.5 in this particular piece. 00:35:03.200 --> 00:35:04.410 align:middle line:90% Who votes yes? 00:35:04.410 --> 00:35:06.526 align:middle line:84% Who votes we might be permanently syncopated? 00:35:06.526 --> 00:35:14.750 align:middle line:90% 00:35:14.750 --> 00:35:18.770 align:middle line:84% Spend more time reading code than writing code. 00:35:18.770 --> 00:35:21.950 align:middle line:84% AUDIENCE: It looks like we might, yeah, because it always 00:35:21.950 --> 00:35:27.470 align:middle line:84% switches the quarter length if we're not on the downbeat. 00:35:27.470 --> 00:35:29.720 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: Oh, did I do that wrong? 00:35:29.720 --> 00:35:32.360 align:middle line:84% Do we want to keep it-- we wanted 00:35:32.360 --> 00:35:35.300 align:middle line:90% to-- do I have a misplaced not? 00:35:35.300 --> 00:35:38.582 align:middle line:84% So that's what we wanted a little bit. 00:35:38.582 --> 00:35:41.220 align:middle line:84% If we're on the downbeat, we can change to a new rhythm. 00:35:41.220 --> 00:35:44.480 align:middle line:90% If we're not, we can't. 00:35:44.480 --> 00:35:45.890 align:middle line:90% I think let's run. 00:35:45.890 --> 00:35:49.193 align:middle line:90% Let's generate this. 00:35:49.193 --> 00:35:51.125 align:middle line:90% AUDIENCE: [INAUDIBLE]. 00:35:51.125 --> 00:35:53.000 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: Yeah, so what's 00:35:53.000 --> 00:35:54.290 align:middle line:90% the consequence of this? 00:35:54.290 --> 00:35:55.597 align:middle line:90% Somebody else. 00:35:55.597 --> 00:35:58.796 align:middle line:90% [MUSICAL NOTES] 00:35:58.796 --> 00:35:59.263 align:middle line:90% 00:35:59.263 --> 00:36:00.680 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: Yep. 00:36:00.680 --> 00:36:03.013 align:middle line:84% AUDIENCE: It's the same type of notes that are changing. 00:36:03.013 --> 00:36:05.013 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: It's the same type 00:36:05.013 --> 00:36:06.377 align:middle line:90% of notes every measure. 00:36:06.377 --> 00:36:06.960 align:middle line:90% You know what? 00:36:06.960 --> 00:36:08.210 align:middle line:90% I'm going to change that over. 00:36:08.210 --> 00:36:10.650 align:middle line:84% Oh, no, no, we can't determine the length of the piece. 00:36:10.650 --> 00:36:13.530 align:middle line:90% So we might have that. 00:36:13.530 --> 00:36:18.480 align:middle line:84% Yeah, every measure will always have the same type of thing. 00:36:18.480 --> 00:36:19.760 align:middle line:90% This is kind of nice for-- 00:36:19.760 --> 00:36:21.500 align:middle line:84% a lot of my junior high etude books 00:36:21.500 --> 00:36:23.820 align:middle line:84% were like this, that you only switched rhythms. 00:36:23.820 --> 00:36:27.450 align:middle line:84% But maybe what's one way that we can switch a little bit better, 00:36:27.450 --> 00:36:28.130 align:middle line:90% more often? 00:36:28.130 --> 00:36:31.680 align:middle line:84% We could just say, we should rename our variables, 00:36:31.680 --> 00:36:34.320 align:middle line:84% but maybe this will feel much more unified. 00:36:34.320 --> 00:36:37.520 align:middle line:84% We'll just do it on every two notes. 00:36:37.520 --> 00:36:40.950 align:middle line:90% [MUSICAL NOTES] 00:36:40.950 --> 00:36:49.280 align:middle line:90% 00:36:49.280 --> 00:36:50.220 align:middle line:90% OK, that's enough. 00:36:50.220 --> 00:36:53.040 align:middle line:84% That's already becoming more unified. 00:36:53.040 --> 00:36:58.570 align:middle line:84% So what are some things that we could hack into this? 00:36:58.570 --> 00:37:03.622 align:middle line:84% We're done for a little bit, but what are some things 00:37:03.622 --> 00:37:05.705 align:middle line:84% that we can make it sound a little bit more human? 00:37:05.705 --> 00:37:10.740 align:middle line:90% 00:37:10.740 --> 00:37:11.490 align:middle line:90% AUDIENCE: Harmony. 00:37:11.490 --> 00:37:12.640 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: We can add harmony. 00:37:12.640 --> 00:37:13.140 align:middle line:90% Good. 00:37:13.140 --> 00:37:15.100 align:middle line:84% We could add random or appropriate. 00:37:15.100 --> 00:37:16.210 align:middle line:90% Probably more appropriate. 00:37:16.210 --> 00:37:16.918 align:middle line:90% Yeah, Jason. 00:37:16.918 --> 00:37:18.960 align:middle line:84% AUDIENCE: We could allow it to occasionally break 00:37:18.960 --> 00:37:21.330 align:middle line:84% these rules by, well, probably by random chance. 00:37:21.330 --> 00:37:23.580 align:middle line:84% MICHAEL SCOTT ASATO CUTHBERT: So certain random chance 00:37:23.580 --> 00:37:24.580 align:middle line:90% that we break the rules. 00:37:24.580 --> 00:37:26.920 align:middle line:84% Yeah, one of the things here, I'm not going to do this. 00:37:26.920 --> 00:37:29.760 align:middle line:84% But I gave a particular random chance 00:37:29.760 --> 00:37:35.400 align:middle line:84% that something might go up an octave in another time 00:37:35.400 --> 00:37:42.210 align:middle line:84% when I did this, and there was some danger in allowing that. 00:37:42.210 --> 00:37:44.890 align:middle line:84% This is a particular problem with MuseScore, 00:37:44.890 --> 00:37:48.570 align:middle line:84% that if you overflow 256, it jumps. 00:37:48.570 --> 00:37:52.860 align:middle line:84% It's unsigned, which is kind of cool that the world of notation 00:37:52.860 --> 00:37:54.450 align:middle line:90% wraps around. 00:37:54.450 --> 00:37:56.550 align:middle line:90% But yeah. 00:37:56.550 --> 00:38:00.930 align:middle line:84% So I obviously was allowing this one random-- 00:38:00.930 --> 00:38:03.990 align:middle line:84% seven is an octave if you're adding to it 00:38:03.990 --> 00:38:05.620 align:middle line:90% to happen a little too often. 00:38:05.620 --> 00:38:08.320 align:middle line:84% So that can be a danger, but it happens. 00:38:08.320 --> 00:38:11.038 align:middle line:90% Thanks for having that. 00:38:11.038 --> 00:38:13.330 align:middle line:84% Yeah, so those are some of the other things you can do. 00:38:13.330 --> 00:38:16.260 align:middle line:84% Maybe we want to sound a little less-- 00:38:16.260 --> 00:38:18.095 align:middle line:90% I'm going to start doing this. 00:38:18.095 --> 00:38:19.720 align:middle line:84% I just need to not make it full screen. 00:38:19.720 --> 00:38:21.450 align:middle line:84% You can make it a little bit less human 00:38:21.450 --> 00:38:24.330 align:middle line:84% and make it do these weird microtones, or not 00:38:24.330 --> 00:38:27.410 align:middle line:84% less human, but less traditional. 00:38:27.410 --> 00:38:29.430 align:middle line:90% And where'd that go? 00:38:29.430 --> 00:38:31.210 align:middle line:90% Here we can hear it. 00:38:31.210 --> 00:38:32.800 align:middle line:90% These, unfortunately-- 00:38:32.800 --> 00:38:36.104 align:middle line:90% [MUSICAL NOTES] 00:38:36.104 --> 00:38:41.310 align:middle line:90% 00:38:41.310 --> 00:38:45.240 align:middle line:84% So there are ways that you can do some things that 00:38:45.240 --> 00:38:48.430 align:middle line:84% allow you to be more experimental if you want, 00:38:48.430 --> 00:38:49.750 align:middle line:90% or less experimental. 00:38:49.750 --> 00:38:52.030 align:middle line:90% Great. 00:38:52.030 --> 00:38:54.410 align:middle line:90% So we'll keep playing with that. 00:38:54.410 --> 00:38:56.230 align:middle line:84% What you'll probably realize, one 00:38:56.230 --> 00:39:00.520 align:middle line:84% of the things we're building up to, so you're given a problem 00:39:00.520 --> 00:39:02.210 align:middle line:90% set 7.8, as I said. 00:39:02.210 --> 00:39:04.250 align:middle line:90% It's not in 7/8 time. 00:39:04.250 --> 00:39:06.670 align:middle line:90% That would be too lopsided. 00:39:06.670 --> 00:39:11.470 align:middle line:84% But what it is, is it gives you some work with scales 00:39:11.470 --> 00:39:12.950 align:middle line:90% and with Roman numerals. 00:39:12.950 --> 00:39:16.730 align:middle line:84% And then there's two parts for you to do an open ended thing. 00:39:16.730 --> 00:39:19.510 align:middle line:84% One is, we're going to be asking you, in the first half 00:39:19.510 --> 00:39:22.360 align:middle line:84% to write a Roman numeral analyzer just 00:39:22.360 --> 00:39:26.200 align:middle line:90% for triads in major. 00:39:26.200 --> 00:39:29.230 align:middle line:84% And it's just going to be for block chords. 00:39:29.230 --> 00:39:30.410 align:middle line:90% I'll give you three notes. 00:39:30.410 --> 00:39:33.110 align:middle line:84% You tell me what Roman numeral it is. 00:39:33.110 --> 00:39:36.090 align:middle line:90% If you're a little bit behind-- 00:39:36.090 --> 00:39:36.590 align:middle line:90% sorry. 00:39:36.590 --> 00:39:39.700 align:middle line:84% If it's been a while since you've done Roman numeral 00:39:39.700 --> 00:39:43.430 align:middle line:84% analysis or you feel a little bit weak on it, collaborate. 00:39:43.430 --> 00:39:45.260 align:middle line:90% Talk with other people on that. 00:39:45.260 --> 00:39:47.890 align:middle line:84% Make sure that you're really clear what a diminished triad 00:39:47.890 --> 00:39:49.880 align:middle line:90% and all that things is. 00:39:49.880 --> 00:39:52.320 align:middle line:84% So grab somebody before you leave the room. 00:39:52.320 --> 00:39:55.070 align:middle line:84% And then there's a little open ended thing 00:39:55.070 --> 00:39:58.020 align:middle line:84% where I just ask you, do something a little bit more. 00:39:58.020 --> 00:40:00.060 align:middle line:84% Maybe you have something that works in minor. 00:40:00.060 --> 00:40:01.950 align:middle line:84% Maybe you have it do it seventh chords. 00:40:01.950 --> 00:40:04.280 align:middle line:84% Maybe you work in jazz and popular music, 00:40:04.280 --> 00:40:06.170 align:middle line:84% and you think Roman numerals, oh, that's 00:40:06.170 --> 00:40:07.530 align:middle line:90% only for classical music. 00:40:07.530 --> 00:40:09.830 align:middle line:84% I'm going to have it do chord symbols, CDM, 00:40:09.830 --> 00:40:11.970 align:middle line:90% and things like that instead. 00:40:11.970 --> 00:40:15.510 align:middle line:84% So whatever you want, there'll be just a small part of it. 00:40:15.510 --> 00:40:16.635 align:middle line:90% I should have it open. 00:40:16.635 --> 00:40:18.260 align:middle line:84% In fact, I think I-- no, I only have it 00:40:18.260 --> 00:40:19.380 align:middle line:90% with all the answers open. 00:40:19.380 --> 00:40:21.245 align:middle line:90% Never mind. 00:40:21.245 --> 00:40:28.130 align:middle line:84% And then, the last part is, I want an algorithmic composition 00:40:28.130 --> 00:40:30.020 align:middle line:90% that you like. 00:40:30.020 --> 00:40:31.590 align:middle line:90% It can be multi-part. 00:40:31.590 --> 00:40:33.780 align:middle line:84% It could be single part if it's interesting. 00:40:33.780 --> 00:40:36.650 align:middle line:84% It could be with chords, without chords, but just 00:40:36.650 --> 00:40:40.440 align:middle line:84% something where you've done some of interesting music theory. 00:40:40.440 --> 00:40:42.680 align:middle line:84% I'll say that what we've given here 00:40:42.680 --> 00:40:45.980 align:middle line:84% is the quintessential D minus so far. 00:40:45.980 --> 00:40:48.990 align:middle line:84% Try to go better than, maybe it's a D. Actually, 00:40:48.990 --> 00:40:51.300 align:middle line:84% but at this point, we're getting a little bit better. 00:40:51.300 --> 00:40:53.670 align:middle line:84% Don't spend the entire semester on it. 00:40:53.670 --> 00:40:56.690 align:middle line:84% This is not meant to be the gigantic project 00:40:56.690 --> 00:40:58.700 align:middle line:90% that you take forever on. 00:40:58.700 --> 00:41:02.150 align:middle line:84% The only restriction I'm giving is 00:41:02.150 --> 00:41:04.850 align:middle line:84% that I'd like it to be about a minute 00:41:04.850 --> 00:41:09.270 align:middle line:84% long, not two minutes long, not 12 minutes long, 00:41:09.270 --> 00:41:11.300 align:middle line:90% because I have to-- 00:41:11.300 --> 00:41:13.890 align:middle line:84% if everybody turns in a 10 minute composition, 00:41:13.890 --> 00:41:16.850 align:middle line:84% then I'll be listening to algorithmic compositions 00:41:16.850 --> 00:41:19.070 align:middle line:90% forever. 00:41:19.070 --> 00:41:22.850 align:middle line:84% So if you can do that, that's going to be that main part. 00:41:22.850 --> 00:41:26.180 align:middle line:84% Any questions about what we did with playing around 00:41:26.180 --> 00:41:28.183 align:middle line:84% with algorithmic compositions today? 00:41:28.183 --> 00:41:35.846 align:middle line:90% 00:41:35.846 --> 00:41:39.500 align:middle line:84% Then I'm going to be a little egotistical for a second 00:41:39.500 --> 00:41:46.500 align:middle line:84% and talk about one of my first algorithmic compositions 00:41:46.500 --> 00:41:50.580 align:middle line:84% that, at some point, people were performing. 00:41:50.580 --> 00:41:52.950 align:middle line:84% And it's a piece for this group called the Bang 00:41:52.950 --> 00:41:55.020 align:middle line:84% on a Can All-Stars, which some of you 00:41:55.020 --> 00:41:58.060 align:middle line:84% might recognize the clarinetist is Evan Ziporyn. 00:41:58.060 --> 00:42:01.390 align:middle line:84% So another professor who's here, who does a lot of things. 00:42:01.390 --> 00:42:04.930 align:middle line:84% And it was based on Markov chains. 00:42:04.930 --> 00:42:07.980 align:middle line:84% So the notion of what I wanted was 00:42:07.980 --> 00:42:12.300 align:middle line:84% some kind of a melody that was a little bit boring-- 00:42:12.300 --> 00:42:14.890 align:middle line:84% not boring, but just undistinguished. 00:42:14.890 --> 00:42:17.650 align:middle line:84% Except it had these weird bumps in it, 00:42:17.650 --> 00:42:19.810 align:middle line:84% some things that sounded a little bit weird, 00:42:19.810 --> 00:42:24.270 align:middle line:84% like this opening tritone, and then 00:42:24.270 --> 00:42:26.010 align:middle line:90% some of these weird gestures. 00:42:26.010 --> 00:42:28.600 align:middle line:84% And so some things that you'd hear enough times, you're like, 00:42:28.600 --> 00:42:31.530 align:middle line:84% OK, I know that that's what the piece is about. 00:42:31.530 --> 00:42:34.067 align:middle line:90% And try to see if this is-- 00:42:34.067 --> 00:42:35.400 align:middle line:90% I can't remember when it's from. 00:42:35.400 --> 00:42:36.460 align:middle line:90% 2005. 00:42:36.460 --> 00:42:39.720 align:middle line:84% It was a little bit pre-deep learning and things, so just 00:42:39.720 --> 00:42:42.210 align:middle line:84% probability theory, whether or not 00:42:42.210 --> 00:42:45.750 align:middle line:84% the Markov chain would do something interesting things 00:42:45.750 --> 00:42:46.390 align:middle line:90% with this. 00:42:46.390 --> 00:42:48.420 align:middle line:84% And then one of the things I really wanted 00:42:48.420 --> 00:42:51.300 align:middle line:84% was someplace like an old record player 00:42:51.300 --> 00:42:56.050 align:middle line:84% that would get stuck in a loop and need to be jumped out of it. 00:42:56.050 --> 00:43:01.050 align:middle line:84% So I have a few places where the same thing happens over, 00:43:01.050 --> 00:43:02.200 align:middle line:90% and over, and over again. 00:43:02.200 --> 00:43:04.075 align:middle line:84% If you're thinking about what we were talking 00:43:04.075 --> 00:43:07.650 align:middle line:84% about at the end of last class about the probability of B, 00:43:07.650 --> 00:43:12.600 align:middle line:84% given A, the next note, or we could say n2, 00:43:12.600 --> 00:43:17.940 align:middle line:84% given n1, this is-- what order or Markov chain 00:43:17.940 --> 00:43:20.250 align:middle line:90% are we talking about here? 00:43:20.250 --> 00:43:22.930 align:middle line:84% First order, based on one previous note. 00:43:22.930 --> 00:43:25.440 align:middle line:84% I used a bunch of different orders in this piece, 00:43:25.440 --> 00:43:28.720 align:middle line:84% depending on how random I wanted it to be. 00:43:28.720 --> 00:43:33.540 align:middle line:84% I wanted that, when you got to that note A, 00:43:33.540 --> 00:43:36.330 align:middle line:84% that the probability, given A, was 00:43:36.330 --> 00:43:40.300 align:middle line:84% very high that you would do another A. 00:43:40.300 --> 00:43:42.980 align:middle line:90% So I weighed the probability. 00:43:42.980 --> 00:43:46.700 align:middle line:84% So the probability of an A, given an A, was, I don't know, 00:43:46.700 --> 00:43:48.770 align:middle line:90% something like 0.9 or something. 00:43:48.770 --> 00:43:51.700 align:middle line:84% So the computer would get stuck in this particular groove 00:43:51.700 --> 00:43:52.665 align:middle line:90% at some points. 00:43:52.665 --> 00:43:55.270 align:middle line:90% 00:43:55.270 --> 00:43:57.980 align:middle line:90% And we'll see. 00:43:57.980 --> 00:44:02.890 align:middle line:84% What I would do at some points is just keep the melody, 00:44:02.890 --> 00:44:06.410 align:middle line:84% just keep the notes, and let the rhythms be out. 00:44:06.410 --> 00:44:12.440 align:middle line:84% But if this is treble clef, that's an opening gesture of B, 00:44:12.440 --> 00:44:15.685 align:middle line:90% F, comes back a lot B, F. 00:44:15.685 --> 00:44:17.080 align:middle line:90% [MUSICAL NOTES] 00:44:17.080 --> 00:44:19.630 align:middle line:84% You'll hear that a lot, and then getting stuck in that 00:44:19.630 --> 00:44:20.600 align:middle line:90% a groove again. 00:44:20.600 --> 00:44:21.910 align:middle line:90% Da, da, da. 00:44:21.910 --> 00:44:24.220 align:middle line:90% You'll hear that. 00:44:24.220 --> 00:44:27.710 align:middle line:84% And what I did was, back in the old days, 00:44:27.710 --> 00:44:31.430 align:middle line:84% I just had it generate a whole bunch of melodies, 00:44:31.430 --> 00:44:32.960 align:middle line:90% and I kept the ones I wanted. 00:44:32.960 --> 00:44:39.130 align:middle line:84% So I wanted to do my computer type of thing. 00:44:39.130 --> 00:44:46.330 align:middle line:84% One last little thing is that I had this sort of progression 00:44:46.330 --> 00:44:51.100 align:middle line:84% that begins one beat per second, so 60 BPM, then moves. 00:44:51.100 --> 00:44:53.930 align:middle line:84% There was a section at 1.5 beats per second, 00:44:53.930 --> 00:44:58.420 align:middle line:84% 2, 3, 4, and then the final note is a dissonance that 00:44:58.420 --> 00:44:59.990 align:middle line:90% vibrates five beats per second. 00:44:59.990 --> 00:45:02.660 align:middle line:84% So that's kind of the little Easter egg in the piece. 00:45:02.660 --> 00:45:05.110 align:middle line:84% And then I took the 1.5 beats per second 00:45:05.110 --> 00:45:09.190 align:middle line:84% and moved it, because that's what composers do. 00:45:09.190 --> 00:45:12.610 align:middle line:90% When the piece was premiered-- 00:45:12.610 --> 00:45:17.150 align:middle line:84% they'll never be happy with me at the printing office again-- 00:45:17.150 --> 00:45:19.750 align:middle line:84% I gave every person in the audience 00:45:19.750 --> 00:45:21.610 align:middle line:84% a different set of program notes that 00:45:21.610 --> 00:45:24.100 align:middle line:84% began by saying the opening of the piece 00:45:24.100 --> 00:45:25.220 align:middle line:90% is rather straightforward. 00:45:25.220 --> 00:45:28.160 align:middle line:84% And then I wrote about 10 pages worth of program notes, 00:45:28.160 --> 00:45:31.580 align:middle line:84% and everyone got a Markov chain of the program notes. 00:45:31.580 --> 00:45:33.160 align:middle line:90% So it has something. 00:45:33.160 --> 00:45:35.590 align:middle line:84% Kind of looks like something, and then it 00:45:35.590 --> 00:45:39.530 align:middle line:84% ended always with a hint of order, no matter what they had. 00:45:39.530 --> 00:45:42.290 align:middle line:84% So I'm just going to be indulgent and play 00:45:42.290 --> 00:45:47.502 align:middle line:84% a little bit of something I wrote a while back. 00:45:47.502 --> 00:45:50.988 align:middle line:90% [MUSIC PLAYING] 00:45:50.988 --> 00:47:18.750 align:middle line:90% 00:47:18.750 --> 00:47:21.215 align:middle line:84% And here's where I let the computer start taking over. 00:47:21.215 --> 00:47:24.100 align:middle line:90% 00:47:24.100 --> 00:47:27.586 align:middle line:90% [MUSIC PLAYING] 00:47:27.586 --> 00:48:06.930 align:middle line:90% 00:48:06.930 --> 00:48:09.540 align:middle line:84% I think I told the computer to multiply the intervals by 2 00:48:09.540 --> 00:48:11.186 align:middle line:90% for more leapy. 00:48:11.186 --> 00:48:14.658 align:middle line:90% [MUSIC PLAYING] 00:48:14.658 --> 00:49:16.750 align:middle line:90% 00:49:16.750 --> 00:49:19.010 align:middle line:84% This is completely Markov chained here. 00:49:19.010 --> 00:49:22.482 align:middle line:90% [MUSIC PLAYING] 00:49:22.482 --> 00:50:16.062 align:middle line:90% 00:50:16.062 --> 00:50:17.895 align:middle line:84% You can kind of hear where it's stuck again. 00:50:17.895 --> 00:50:21.272 align:middle line:90% [MUSIC PLAYING] 00:50:21.272 --> 00:51:00.080 align:middle line:90% 00:51:00.080 --> 00:51:03.090 align:middle line:84% Any time you see a two measure repeat, the timing didn't work. 00:51:03.090 --> 00:51:06.555 align:middle line:90% [MUSIC PLAYING] 00:51:06.555 --> 00:51:56.120 align:middle line:90% 00:51:56.120 --> 00:52:00.590 align:middle line:90% That's 2002. 00:52:00.590 --> 00:52:02.400 align:middle line:90% Well, thanks, everybody. 00:52:02.400 --> 00:52:04.227 align:middle line:84% You don't need to write that much. 00:52:04.227 --> 00:52:06.810 align:middle line:84% Please don't write that much for your algorithmic composition, 00:52:06.810 --> 00:52:11.330 align:middle line:84% but feel free to have a mixture of your own personality 00:52:11.330 --> 00:52:14.810 align:middle line:84% and your own desires, and let the computer do some things 00:52:14.810 --> 00:52:17.680 align:middle line:90% that kind of inspires you. 00:52:17.680 --> 00:52:28.000 align:middle line:90%