WEBVTT

00:00:00.000 --> 00:00:04.473 align:middle line:84%
[SQUEAKING]
[RUSTLING] [CLICKING]

00:00:04.473 --> 00:00:09.950 align:middle line:90%


00:00:09.950 --> 00:00:12.500 align:middle line:84%
MICHAEL CUTHBERT: Hello,
OpenCourseWare learners.

00:00:12.500 --> 00:00:14.730 align:middle line:84%
This is Professor
Cuthbert talking to you.

00:00:14.730 --> 00:00:16.850 align:middle line:84%
And I just wanted
to say a little bit

00:00:16.850 --> 00:00:19.550 align:middle line:84%
about why the next
two classes are

00:00:19.550 --> 00:00:23.390 align:middle line:90%
not to be found on this site.

00:00:23.390 --> 00:00:26.000 align:middle line:84%
MIT and the music
department was very

00:00:26.000 --> 00:00:29.210 align:middle line:84%
fortunate to be able
to hire a new professor

00:00:29.210 --> 00:00:32.009 align:middle line:84%
in computational
music that semester.

00:00:32.009 --> 00:00:34.700 align:middle line:84%
And one of the things that
we value as a department

00:00:34.700 --> 00:00:37.500 align:middle line:84%
is people's ability
to be good teachers.

00:00:37.500 --> 00:00:41.420 align:middle line:84%
So as part of their
interview, the candidates

00:00:41.420 --> 00:00:44.360 align:middle line:84%
came and took over
one of the classes

00:00:44.360 --> 00:00:47.900 align:middle line:84%
to teach the
content of the class

00:00:47.900 --> 00:00:50.160 align:middle line:84%
and a little bit of
their specialties, too.

00:00:50.160 --> 00:00:55.400 align:middle line:84%
And two of those classes that
were used were this class,

00:00:55.400 --> 00:00:59.270 align:middle line:90%
21m.383.

00:00:59.270 --> 00:01:02.870 align:middle line:84%
Both of the candidates, I
can say, did amazing work,

00:01:02.870 --> 00:01:05.780 align:middle line:84%
and I wish I could share
them their videos with you.

00:01:05.780 --> 00:01:09.730 align:middle line:84%
But we have a feeling that
it's already stressful enough

00:01:09.730 --> 00:01:13.750 align:middle line:84%
to be doing a job interview
and to have your teaching being

00:01:13.750 --> 00:01:17.860 align:middle line:84%
evaluated by people in
the room, and then to know

00:01:17.860 --> 00:01:21.670 align:middle line:84%
that whatever things you did
in an unfamiliar classroom

00:01:21.670 --> 00:01:25.820 align:middle line:84%
with unfamiliar students
would also be put out online.

00:01:25.820 --> 00:01:27.350 align:middle line:90%
It's just a little bit too much.

00:01:27.350 --> 00:01:30.230 align:middle line:84%
So these-- to ask,
too much to ask.

00:01:30.230 --> 00:01:32.420 align:middle line:84%
So these videos
are not available.

00:01:32.420 --> 00:01:34.220 align:middle line:84%
I want to say a
little bit, though,

00:01:34.220 --> 00:01:36.700 align:middle line:90%
what was in each of them.

00:01:36.700 --> 00:01:40.870 align:middle line:84%
The first class was
the continuation

00:01:40.870 --> 00:01:45.200 align:middle line:84%
of our series on music,
cognition, and computation,

00:01:45.200 --> 00:01:50.680 align:middle line:84%
and how computers can aid
the process of understanding

00:01:50.680 --> 00:01:53.740 align:middle line:84%
how the brain processes
music, and also,

00:01:53.740 --> 00:01:57.880 align:middle line:84%
how historically, and
I think in the future,

00:01:57.880 --> 00:02:02.710 align:middle line:84%
insights from music cognition
have improved the things

00:02:02.710 --> 00:02:04.570 align:middle line:90%
that we do on the computer.

00:02:04.570 --> 00:02:07.670 align:middle line:84%
The most important
example of this

00:02:07.670 --> 00:02:15.440 align:middle line:84%
is Carol Krumhansl's
probe tone experiments

00:02:15.440 --> 00:02:22.190 align:middle line:84%
in understanding musical
expectation in scales and keys.

00:02:22.190 --> 00:02:25.490 align:middle line:84%
And these experiments,
which help

00:02:25.490 --> 00:02:30.860 align:middle line:84%
to understand how each
note plays in our ability

00:02:30.860 --> 00:02:33.290 align:middle line:84%
to understand what key
we're in at this point

00:02:33.290 --> 00:02:38.300 align:middle line:84%
underlies quite a number of
algorithms being used today

00:02:38.300 --> 00:02:42.750 align:middle line:84%
by computers to automatically
detect the keys of compositions.

00:02:42.750 --> 00:02:47.910 align:middle line:84%
And this includes Music 21's
default analyze key method.

00:02:47.910 --> 00:02:53.780 align:middle line:84%
And this is why the assignments
on understanding key analysis

00:02:53.780 --> 00:03:01.130 align:middle line:84%
went hand in hand with
the section on cognition.

00:03:01.130 --> 00:03:03.800 align:middle line:84%
The lecture also
covered and gave

00:03:03.800 --> 00:03:07.140 align:middle line:84%
the opportunity to discuss
some of the experiments

00:03:07.140 --> 00:03:09.840 align:middle line:84%
reported by David
Huron in his book

00:03:09.840 --> 00:03:12.990 align:middle line:84%
Sweet Anticipation,
of which the students

00:03:12.990 --> 00:03:16.050 align:middle line:90%
were assigned to do readings.

00:03:16.050 --> 00:03:18.030 align:middle line:84%
I would encourage,
as a substitute

00:03:18.030 --> 00:03:21.480 align:middle line:84%
for the content of
this lecture, for you

00:03:21.480 --> 00:03:29.820 align:middle line:84%
to look deeply at this
book by a prominent music

00:03:29.820 --> 00:03:34.050 align:middle line:84%
cognition and expert, and
also one of the breakthrough

00:03:34.050 --> 00:03:39.000 align:middle line:84%
founders of our field of
computational music theory

00:03:39.000 --> 00:03:42.150 align:middle line:90%
and musicology, David Huron.

00:03:42.150 --> 00:03:44.910 align:middle line:84%
And doing so, I think,
would give a good sense

00:03:44.910 --> 00:03:48.120 align:middle line:84%
of the majority of
what was contained

00:03:48.120 --> 00:03:53.400 align:middle line:84%
in the lecture beyond
what the candidate spoke

00:03:53.400 --> 00:03:55.726 align:middle line:84%
about her own
particular expertise.

00:03:55.726 --> 00:03:59.400 align:middle line:90%


00:03:59.400 --> 00:04:02.880 align:middle line:84%
The second guest lecturer
covered the topic

00:04:02.880 --> 00:04:08.130 align:middle line:84%
of the history of algorithmic
composition from its beginnings,

00:04:08.130 --> 00:04:11.220 align:middle line:84%
even before the
computer, until today.

00:04:11.220 --> 00:04:17.579 align:middle line:84%
And this lecturer covered the
main reading for this week,

00:04:17.579 --> 00:04:23.330 align:middle line:84%
which is on the "Illiac Suite,"
one of the first pieces to be

00:04:23.330 --> 00:04:26.930 align:middle line:84%
composed by computers using
algorithms and to generate

00:04:26.930 --> 00:04:32.430 align:middle line:84%
notes, and continuing up till
today with advances in AI,

00:04:32.430 --> 00:04:39.230 align:middle line:84%
such as the music transformer
by Ana Huang at Google and other

00:04:39.230 --> 00:04:41.220 align:middle line:90%
new developments.

00:04:41.220 --> 00:04:44.630 align:middle line:84%
One of the reasons
why it's so important

00:04:44.630 --> 00:04:50.930 align:middle line:84%
to look at algorithmic music
composition in a broader

00:04:50.930 --> 00:04:53.570 align:middle line:84%
historical scope
is it's too easy

00:04:53.570 --> 00:04:56.990 align:middle line:84%
to get hung up on the
particular problem

00:04:56.990 --> 00:04:59.690 align:middle line:84%
that we have today, where
we're getting closer

00:04:59.690 --> 00:05:03.800 align:middle line:84%
and closer to having the
computer produce musical scores

00:05:03.800 --> 00:05:07.500 align:middle line:84%
that sound like
other compositions,

00:05:07.500 --> 00:05:11.940 align:middle line:84%
like by particular composers,
and to work on that problem

00:05:11.940 --> 00:05:13.450 align:middle line:90%
without asking.

00:05:13.450 --> 00:05:14.800 align:middle line:90%
well, why?

00:05:14.800 --> 00:05:16.240 align:middle line:90%
Why should we care?

00:05:16.240 --> 00:05:21.480 align:middle line:84%
Why do we want the computer
to compose a new piece

00:05:21.480 --> 00:05:26.200 align:middle line:84%
in the style of Johann Sebastian
Bach, unless, by chance,

00:05:26.200 --> 00:05:29.760 align:middle line:84%
you've already heard every piece
by Bach, and you're just bored

00:05:29.760 --> 00:05:33.270 align:middle line:84%
and you need more,
which most of us

00:05:33.270 --> 00:05:37.080 align:middle line:84%
haven't explored what's
actually out there?

00:05:37.080 --> 00:05:41.080 align:middle line:84%
Instead, by looking back
at the earliest history,

00:05:41.080 --> 00:05:44.400 align:middle line:84%
when having a computer
actually write music

00:05:44.400 --> 00:05:49.380 align:middle line:84%
at that level of sophistication
was so far from being possible

00:05:49.380 --> 00:05:54.420 align:middle line:84%
and looking at what people
wanted to do with the computer,

00:05:54.420 --> 00:05:56.860 align:middle line:90%
it helps to answer questions.

00:05:56.860 --> 00:06:03.150 align:middle line:84%
How can the computer be
used to come up with ideas

00:06:03.150 --> 00:06:05.850 align:middle line:84%
or take ideas that I
have that will take too

00:06:05.850 --> 00:06:08.220 align:middle line:84%
long for me to
generate on my own

00:06:08.220 --> 00:06:11.110 align:middle line:84%
and make that part
more efficient?

00:06:11.110 --> 00:06:17.840 align:middle line:84%
How can it be a tool to help
me be a better human composer,

00:06:17.840 --> 00:06:20.870 align:middle line:90%
or just better human?

00:06:20.870 --> 00:06:25.730 align:middle line:84%
So the class went through some
of the earliest algorithmic

00:06:25.730 --> 00:06:28.830 align:middle line:84%
compositions from
Guido of Arezzo,

00:06:28.830 --> 00:06:31.880 align:middle line:84%
a monk living in
the 11th century,

00:06:31.880 --> 00:06:34.850 align:middle line:84%
through Mozart's
musical dice games,

00:06:34.850 --> 00:06:39.140 align:middle line:84%
up until the "Illiac
Suite" and then beyond.

00:06:39.140 --> 00:06:41.730 align:middle line:84%
Some of what's
covered in this class

00:06:41.730 --> 00:06:45.080 align:middle line:84%
also appears in
the various videos

00:06:45.080 --> 00:06:48.530 align:middle line:84%
that accompany this
class that weren't

00:06:48.530 --> 00:06:55.730 align:middle line:84%
shown to students this year,
but are available for you.

00:06:55.730 --> 00:07:00.290 align:middle line:84%
And other aspects touched
on the guest speakers

00:07:00.290 --> 00:07:04.500 align:middle line:84%
particular expertise, which was
very valuable to be brought in.

00:07:04.500 --> 00:07:07.910 align:middle line:84%
Anybody else teaching
this class would

00:07:07.910 --> 00:07:11.750 align:middle line:84%
bring in her or his own
specialties that would make

00:07:11.750 --> 00:07:13.800 align:middle line:90%
that class different from mine.

00:07:13.800 --> 00:07:15.750 align:middle line:84%
And nobody should
teach this class

00:07:15.750 --> 00:07:20.490 align:middle line:84%
exactly the same way I would
teach it, or have taught it.

00:07:20.490 --> 00:07:24.880 align:middle line:84%
So that gives a little bit of
a nutshell of what's happening.

00:07:24.880 --> 00:07:29.370 align:middle line:84%
The students then went on to
write their own algorithms

00:07:29.370 --> 00:07:32.890 align:middle line:84%
to generate compositions
algorithmically,

00:07:32.890 --> 00:07:39.330 align:middle line:84%
and they did a very good job of
avoiding the problem of trying

00:07:39.330 --> 00:07:43.080 align:middle line:84%
to get the computer to do
something that we already

00:07:43.080 --> 00:07:46.890 align:middle line:84%
do better, that we are
not interested really

00:07:46.890 --> 00:07:53.700 align:middle line:84%
in the computer's output,
and that we just would only

00:07:53.700 --> 00:07:57.670 align:middle line:84%
want to listen to if we
knew a human created.

00:07:57.670 --> 00:07:59.460 align:middle line:84%
And instead, they
went beyond that

00:07:59.460 --> 00:08:01.440 align:middle line:84%
and created a
piece of music that

00:08:01.440 --> 00:08:05.520 align:middle line:84%
were interesting in themselves
and that no human would ever

00:08:05.520 --> 00:08:06.485 align:middle line:90%
choose to write.

00:08:06.485 --> 00:08:10.680 align:middle line:90%


00:08:10.680 --> 00:08:16.320 align:middle line:84%
In addition to the two classes
that job candidates covered,

00:08:16.320 --> 00:08:22.970 align:middle line:84%
two other classes at around
this time needed to be removed

00:08:22.970 --> 00:08:26.280 align:middle line:84%
and the contents pushed
into other classes

00:08:26.280 --> 00:08:33.380 align:middle line:84%
because MIT was in the process
of proposing to the full faculty

00:08:33.380 --> 00:08:35.340 align:middle line:90%
a new master's program--

00:08:35.340 --> 00:08:37.220 align:middle line:90%
two new master's programs--

00:08:37.220 --> 00:08:40.500 align:middle line:84%
in music, technology
and computation.

00:08:40.500 --> 00:08:44.270 align:middle line:84%
And it happened
that these proposals

00:08:44.270 --> 00:08:46.650 align:middle line:84%
had to happen at the
same time as this class.

00:08:46.650 --> 00:08:50.130 align:middle line:84%
And so a little bit of
time was lost on that.

00:08:50.130 --> 00:08:53.665 align:middle line:84%
I'm hoping that if you've
been interested in this class

00:08:53.665 --> 00:08:55.040 align:middle line:84%
and want to go
forward, you might

00:08:55.040 --> 00:08:57.800 align:middle line:84%
look into these
programs, which I

00:08:57.800 --> 00:09:03.050 align:middle line:84%
assume will be live at the time
that these videos are released.

00:09:03.050 --> 00:09:04.640 align:middle line:84%
I think that this
is all that I'm

00:09:04.640 --> 00:09:07.130 align:middle line:84%
going to be able to
provide specially

00:09:07.130 --> 00:09:10.320 align:middle line:84%
for the OpenCourseWare
audience in this class,

00:09:10.320 --> 00:09:14.210 align:middle line:84%
except at the very end, where
I have a little message for you

00:09:14.210 --> 00:09:16.050 align:middle line:90%
that I recorded at the time.

00:09:16.050 --> 00:09:18.130 align:middle line:90%
Thank you.

00:09:18.130 --> 00:09:21.000 align:middle line:90%