WEBVTT
00:00:03.800 --> 00:00:07.560
I started listening to Lo-Fi to figure out if
there's any causation behind that
00:00:07.560 --> 00:00:12.200
correlation. I tried to teach an AI to
generate the ideal study music.
00:00:12.440 --> 00:00:16.240
I feel like you could listen to it if the
volume was at like one,
00:00:16.320 --> 00:00:19.160
and then you walked away and you listened to
different music.
00:00:19.320 --> 00:00:20.720
Please don't release it.
00:00:20.800 --> 00:00:25.440
Let me explain. Thank you to The Great
Courses Plus for sponsoring this video.
00:00:26.160 --> 00:00:28.680
When it comes down to focus,
I have two modes.
00:00:29.120 --> 00:00:33.200
I either start a task,
blink, and suddenly it's six hours later and
00:00:33.200 --> 00:00:40.920
I skip two meals, or I surround myself with a
controlled level of distraction.
00:00:41.720 --> 00:00:44.720
many different audio playing What are you doing? I'm focusing.
00:00:45.320 --> 00:00:53.680
That was until I found lo-fi hip-hop chill
beats to study/relax to sparkle emoji.
00:00:53.800 --> 00:00:56.760
The titles are a bit much,
but you know what? I'm talking about,
00:00:56.760 --> 00:01:01.828
this girl. Anyway, I started listening to it
back in 2018 and I saw my GPA just explosion
00:01:01.828 --> 00:01:05.230
sound] and sure, there are several reasons as
to why that could have happened,
00:01:05.230 --> 00:01:09.630
but Lo-Fi music definitely makes me feel like
I'm better at focusing.
00:01:09.910 --> 00:01:13.230
Not every song, some slap a little too hard
and pull me right out of the zone,
00:01:13.230 --> 00:01:15.430
but there is a sweet spot.
00:01:15.510 --> 00:01:19.430
So today in this video,
you and I are going to figure out why Lo-Fi
00:01:19.470 --> 00:01:21.590
Hip Hop might help you focus.
00:01:21.590 --> 00:01:25.530
And if we can teach an AI to generate the
perfect mix.
00:01:31.310 --> 00:01:35.750
So I started my research by looking into the
history of Lo-Fi Hip hop.
00:01:36.070 --> 00:01:39.470
After that, I got a bit more familiar with
other experiments people have done with
00:01:39.470 --> 00:01:43.470
machine learning to generate new music,
and I've got two big takeaways from my
00:01:43.470 --> 00:01:47.670
initial readings. The first has to do with
the structure of Lo-Fi Hip Hop.
00:01:48.030 --> 00:01:51.310
Its iconic sound is composed of three parts.
00:01:51.310 --> 00:01:54.910
The first, a jazzy instrumental,
the second,
00:01:54.910 --> 00:01:57.990
a drum beat combo of bass and snares called
Boom BAP.
00:01:58.310 --> 00:02:03.070
It's played intentionally off kilter in a
style pioneered by the producer and rapper J
00:02:03.070 --> 00:02:08.340
Dilla. And finally, intentional environmental
noise that leans even further into the
00:02:08.340 --> 00:02:10.700
imperfections that define the genre.
00:02:11.260 --> 00:02:16.420
It usually plays at around 70 to 95 beats per
minute and falls within this range of
00:02:16.420 --> 00:02:20.700
frequencies. And that gives me a few key
characteristics to look into when trying to
00:02:20.700 --> 00:02:25.740
figure out why Lo-Fi can help people focus,
but it also leads to a second thing.
00:02:25.740 --> 00:02:29.740
It's an ingredient list and recipe for the
things I need to teach my machine.
00:02:29.820 --> 00:02:33.180
Here's the plan. Okay,
here's how I'm gonna make music with a
00:02:33.180 --> 00:02:36.500
machine. Step one I need a bunch of
instrumental and drum midis.
00:02:36.500 --> 00:02:40.180
These are basically files that contain
information about the note that's played, how
00:02:40.180 --> 00:02:42.100
long it's played, a bunch of other stuff.
00:02:42.260 --> 00:02:43.540
I don't know much about it.
00:02:43.540 --> 00:02:45.100
I'm an amateur. Step two.
00:02:45.500 --> 00:02:48.740
I'm gonna sacrifice it to the machine. I'm
going to be using an LSTM.
00:02:48.740 --> 00:02:52.020
So basically you have a sequence of notes.
And then the next step in the sequence you
00:02:52.020 --> 00:02:55.060
tell the machine, "hey,
get really good at guessing what the next
00:02:55.060 --> 00:02:58.540
step is going to be," and then it gets really
good at it.
00:02:58.540 --> 00:03:03.740
Finally, I'm going to smash that drum and
instrumental track together with some ambient
00:03:03.740 --> 00:03:06.730
music, and I put two step twos.
00:03:06.970 --> 00:03:10.250
This is step three. And then profit.
00:03:10.370 --> 00:03:13.690
Sounds easy right? It was not easy.
00:03:14.770 --> 00:03:19.330
I spent the next week swinging between
wanting to throw my computer out a window and
00:03:19.330 --> 00:03:21.370
wanting to throw myself out a window.
00:03:21.610 --> 00:03:24.170
Oh my God. Oh, jeez. Oh my God.
00:03:24.210 --> 00:03:27.170
Oh, Jesus. So it's been like three days.
00:03:27.250 --> 00:03:30.930
My code has run fully without error for the
first time.
00:03:30.930 --> 00:03:35.610
So as long as there's any sound,
I'm gonna be happy.
00:03:35.650 --> 00:03:36.970
Okay, let's just play this.
00:03:46.690 --> 00:03:49.970
I'm not happy. Okay, let's fix this.
00:03:57.410 --> 00:04:01.370
Okay, so I think I've stopped it from just
repeating the same note.
00:04:01.370 --> 00:04:05.610
Basically, we get caught up in this loop
where it was just like "the last note was a
00:04:05.610 --> 00:04:10.730
C. The next note probably going to be a C,"
but then if it thinks that the if the last
00:04:10.730 --> 00:04:13.330
note C, then the next,
it just gets caught up in this loop.
00:04:13.370 --> 00:04:16.170
Instead it has like this short list of likely
next notes.
00:04:16.170 --> 00:04:18.730
And then it randomly samples from that short
list.
00:04:18.850 --> 00:04:33.380
Anyway, here's how it sounds. Janky out of tune single notes
00:04:33.690 --> 00:04:36.090
That hurt my forehead.
00:04:36.130 --> 00:04:37.090
Isn't al.
00:04:38.770 --> 00:04:43.010
All right? That was headed to a dark place,
so I'm going on a walk to calm down.
00:04:43.530 --> 00:04:45.450
I'm also going to do a bit of reading as to
why,
00:04:45.490 --> 00:04:51.690
like, those specific characteristics of Lo-Fi
music might help with studying and focus.
00:04:52.170 --> 00:04:56.170
I'm also going to collect a bit of
environmental noise for the final track.
00:04:57.650 --> 00:05:01.810
The more I read, the more I realized that the
characteristics that make lo fi,
00:05:01.850 --> 00:05:05.130
lo fi, that steady drumbeat,
that frequency range,
00:05:05.130 --> 00:05:09.840
that faux vinyl crackle,
all those things work together to create a
00:05:09.840 --> 00:05:12.800
sonic environment tuned for focus.
00:05:14.160 --> 00:05:17.920
It's really interesting. I should warn you
that music's impact on your mind isn't
00:05:17.920 --> 00:05:21.520
universal. It depends on the task at hand and
the type of person you are.
00:05:21.640 --> 00:05:23.960
So if what you're doing requires a lot of
thinking,
00:05:23.960 --> 00:05:27.400
or you're currently in a place where you're
going to get easily distracted, you should
00:05:27.400 --> 00:05:30.960
probably hit mute. We have a limited capacity
for attention,
00:05:30.960 --> 00:05:33.640
and music uses up some of that essential
resource.
00:05:33.720 --> 00:05:38.320
In reality, if you want to focus,
silence is almost always the better option.
00:05:38.360 --> 00:05:42.440
However, background music can provide two key
benefits.
00:05:42.440 --> 00:05:45.560
So here's how Lo-Fi can help you focus.
00:05:45.800 --> 00:05:50.960
First, it can block out more intrusive noises
by forming an aural cocoon that tunes out or
00:05:50.960 --> 00:05:55.520
reduces distractions,
by anchoring them to a predictable rhythm.
00:05:55.560 --> 00:05:57.800
Second, it can serve as a stimulant.
00:05:57.960 --> 00:06:02.040
Songs that vibe in the fine line between
putting you to sleep and popping off can
00:06:02.040 --> 00:06:04.920
improve your productivity without being too
distracting.
00:06:05.160 --> 00:06:07.920
And that is where Lo-Fi Hip Hop comes in.
00:06:07.960 --> 00:06:11.830
The jazzy instrumentals and drum loops are
refreshing and fun,
00:06:11.950 --> 00:06:16.230
but slow and repetitive enough that they
become subconsciously predictable.
00:06:16.350 --> 00:06:21.110
In addition, the dampened frequencies and
environmental noise act as a blanket that
00:06:21.110 --> 00:06:25.750
blends the music together and pushes it into
the background using spatialization.
00:06:25.910 --> 00:06:28.110
This makes the music sound less like a
recording,
00:06:28.150 --> 00:06:32.230
playing directly in your ears,
and more like something in a physical space.
00:06:32.830 --> 00:06:35.630
Finally, the genre's obsession with
comfortable,
00:06:35.630 --> 00:06:41.150
nostalgic elements like vinyl crackle in
rainy days can help bring listeners peace and
00:06:41.150 --> 00:06:46.750
improve their mood. All of this combines to
make certain lo fi hip hop tracks easy to
00:06:46.790 --> 00:06:49.750
process and ideal for focus.
00:06:49.750 --> 00:06:53.750
And with all of that in mind,
I now think that I know which parts of lo fi
00:06:53.750 --> 00:06:57.550
I should lean into while making new tracks
and which ones I should avoid.
00:06:57.670 --> 00:07:03.310
Now I just need to get the steaming pile of
garbage I call my code to actually function.
00:07:03.710 --> 00:07:09.390
Let's do this. It took another two days and
just endless emotional turmoil,
00:07:09.630 --> 00:07:13.860
but I eventually was ready for my album
release.
00:07:15.420 --> 00:07:16.940
This might have been a bad idea.
00:07:18.940 --> 00:07:20.260
Do you all know why you're here?
00:07:20.460 --> 00:07:22.660
I have no idea what I'm doing here.
00:07:22.900 --> 00:07:24.540
Apparently, this is the listening party.
00:07:24.980 --> 00:07:27.860
Before we get to the songs,
I do have a question for you guys.
00:07:27.900 --> 00:07:29.420
Okay. You know me pretty well.
00:07:30.220 --> 00:07:35.220
On a scale of 1 to 10,
how musical do you think I am?
00:07:35.820 --> 00:07:36.700
Like a three.
00:07:37.660 --> 00:07:38.660
Which one's bad?
00:07:39.260 --> 00:07:39.820
Zero.
00:07:40.420 --> 00:07:41.540
Zero.
00:07:42.380 --> 00:07:43.860
On a scale of 1 to 10.
00:07:44.380 --> 00:07:45.060
Zero.
00:07:45.380 --> 00:07:47.940
You didn't know who Alicia Keys was?
00:07:47.980 --> 00:07:50.020
I was just gonna bring that up.
00:07:50.020 --> 00:07:54.220
So I would like to present to you the music.
00:07:54.300 --> 00:07:56.100
Which track would you guys like to listen to?
00:07:56.380 --> 00:07:57.420
The worst one.
00:07:57.460 --> 00:07:58.620
Yeah. Let's work our way up.
00:08:07.660 --> 00:08:09.220
This is an anxiety machine.
00:08:09.580 --> 00:08:13.970
This sounds like, Now,
I'm not going to compare it to any music.
00:08:13.970 --> 00:08:15.450
That would be a disrespect to music.
00:08:15.810 --> 00:08:19.410
So you're telling me that isn't the lo fi hip
hop beat to chill and relax to?
00:08:19.450 --> 00:08:20.730
A little far from that.
00:08:21.090 --> 00:08:22.570
The next one. Are you ready?
00:08:23.010 --> 00:08:24.090
Which one would you like?
00:08:24.690 --> 00:08:26.290
Wait. Maybe this is the worst one.
00:08:27.730 --> 00:08:28.210
Okay.
00:08:30.810 --> 00:08:35.410
Okay. Wait for this drop.
00:08:36.210 --> 00:08:37.450
There is no drop.
00:08:38.610 --> 00:08:42.490
No. Is this gonna be available in our
newsletter?
00:08:42.610 --> 00:08:43.410
It will be.
00:08:43.730 --> 00:08:45.690
Where can I subscribe to the newsletter?
00:08:45.690 --> 00:08:46.370
Sabrina.
00:08:46.770 --> 00:08:50.890
You can go to
answerinprogress.com/newsletter.
00:08:50.890 --> 00:08:53.290
To sign up. All right. Let's go with the only
good track.
00:08:53.330 --> 00:08:58.450
So to set the expectations this is the best I
was able to do.
00:08:59.050 --> 00:08:59.290
Okay.
00:08:59.330 --> 00:09:01.010
I'm nervous again actually.
00:09:01.050 --> 00:09:08.000
Yeah that waveform is really aggressive. music of mostly kicks and occasional piano notes
00:09:08.050 --> 00:09:10.390
Kind of minecrafty. music continues
00:09:10.390 --> 00:09:16.280
Very elevator music. Like I'm waiting for my
floor just past the seventh floor.
00:09:17.560 --> 00:09:18.360
Yeah.
00:09:18.400 --> 00:09:19.680
Now we're at the 24th floor.
00:09:19.720 --> 00:09:20.400
That was really funny.
00:09:20.520 --> 00:09:22.560
Damn, that elevator's real fast.
00:09:22.680 --> 00:09:23.960
I'm gonna close my eyes.
00:09:28.400 --> 00:09:31.280
It's getting creepier. I think the more that
I listen to it,
00:09:31.280 --> 00:09:34.640
I'm like, I should not listen to this
anymore.
00:09:34.920 --> 00:09:35.800
It's unsettling.
00:09:35.920 --> 00:09:37.080
It is slightly unsettling.
00:09:37.080 --> 00:09:39.800
However, you know, it's just like,
I don't know if it's a chill,
00:09:39.800 --> 00:09:46.680
lo fi beat, but from listening to this,
compared to the first lo fi beat singular
00:09:46.680 --> 00:09:49.200
that you created, this is this is a lot.
00:09:49.240 --> 00:09:50.600
This is a lot of improvement here.
00:09:50.720 --> 00:09:54.320
Here's what it sounds like. If I just use
like an actual drum loop designed by a human
00:09:54.320 --> 00:09:54.680
being.
00:10:00.200 --> 00:10:06.200
This one's good. The big question,
Taha, could you study to this?
00:10:06.640 --> 00:10:08.560
I would definitely study to this.
00:10:08.600 --> 00:10:10.600
I feel like I have studied to this.
00:10:10.720 --> 00:10:11.920
How do you guys think I did?
00:10:12.440 --> 00:10:15.970
I think you did Badly.
00:10:16.020 --> 00:10:17.430
[Laughs]
00:10:17.440 --> 00:10:23.000
I think that the drum loop that you got did a
lot of the heavy lifting here.
00:10:23.080 --> 00:10:24.000
Melissa?
00:10:24.320 --> 00:10:25.650
[Laughs]
00:10:25.960 --> 00:10:28.240
Damn, dude. Speechless.
00:10:29.040 --> 00:10:32.640
In fairness, it's it's pretty impressive that
you got,
00:10:33.160 --> 00:10:35.880
like, some semblance of,
like, vibes.
00:10:36.040 --> 00:10:40.640
Yeah. We've created a new genre of music
beats to be anxious to.
00:10:40.640 --> 00:10:43.280
To beats to be anxious to.
00:10:44.880 --> 00:10:47.960
Okay, so while that wasn't the hardest video
I've ever had to make,
00:10:47.960 --> 00:10:51.800
I definitely felt like I bit off more than I
could chew.
00:10:51.840 --> 00:10:57.160
Way more than once. It was the first time in
a while that I truly didn't know how to even
00:10:57.160 --> 00:11:00.880
start. Every step I just felt out of my
element and I felt like there was just so
00:11:00.880 --> 00:11:02.400
much more I needed to learn.
00:11:02.400 --> 00:11:05.240
But even though it took way longer than
expected,
00:11:05.240 --> 00:11:07.640
it was really nice to go from this.
00:11:10.720 --> 00:11:11.360
To this.
00:11:11.960 --> 00:11:13.480
This one's good.
00:11:13.880 --> 00:11:15.880
Knowing all of the hard work that went behind
it,
00:11:15.880 --> 00:11:20.870
and by the end, we were able to figure out
why lofi hip hop and ambient music can help
00:11:20.870 --> 00:11:25.750
you study and also make a few decent beats of
our own.
00:11:26.110 --> 00:11:30.270
Kind of. Okay, I'll admit it was far from
perfect,
00:11:30.470 --> 00:11:35.750
but if we gave up because our first attempt
or our first dozen attempts weren't perfect,
00:11:35.950 --> 00:11:37.150
we wouldn't get very far.
00:11:37.310 --> 00:11:41.990
You know, that's just a lesson in lo fi and
in life.
00:11:42.310 --> 00:11:45.190
I know that was a bad one,
bye.
00:11:46.390 --> 00:11:47.990
Hey there, I hope you liked that video.
00:11:47.990 --> 00:11:51.870
If you did, please consider sharing it with a
few friends. It is easily the biggest way you
00:11:51.870 --> 00:11:53.030
can help this channel grow.
00:11:53.150 --> 00:11:57.350
Another way is to stick around as I thank the
Great Courses Plus for sponsoring this video
00:11:57.390 --> 00:12:00.390
while I turn myself into a lo fi hip hop
girl.
00:12:00.390 --> 00:12:03.550
So I found that learning things gets a lot
harder after you leave,
00:12:03.550 --> 00:12:06.030
like the guided learning environment of
university.
00:12:06.150 --> 00:12:09.430
And unfortunately I don't have the time or
money to head back.
00:12:10.030 --> 00:12:12.270
Luckily I use the Great Courses Plus.
00:12:12.550 --> 00:12:16.310
It's a subscription on demand video learning
service with these amazing lectures and
00:12:16.310 --> 00:12:20.860
courses from top from top professors from Ivy
League universities and experts from major
00:12:20.860 --> 00:12:25.020
institutions. Your subscription gets you
access to over 11,000 videos covering
00:12:25.020 --> 00:12:27.580
everything from music theory to math and
more.
00:12:27.700 --> 00:12:29.300
I recommend music and the brain.
00:12:29.420 --> 00:12:33.740
It explores the neuroscience of music and
what happens when melody meets the mind.
00:12:33.780 --> 00:12:37.940
It helped me better understand the concepts
that tie Lo-Fi and focus together.
00:12:38.100 --> 00:12:42.020
You can learn at your own pace,
at any time or any place without worrying
00:12:42.020 --> 00:12:44.460
about tests, which is the greatest feeling in
the world.
00:12:44.540 --> 00:12:47.420
You can get started with The Great Courses
Plus today if you head on over to
00:12:47.420 --> 00:12:51.100
thegreatcoursesplus.com/answerinprogress.
00:12:51.100 --> 00:12:53.980
Or you can click the first link in the
description. So get started with the Great
00:12:53.980 --> 00:12:57.580
Courses Plus today. But either way,
have a lovely day.
00:12:57.930 --> 00:12:59.390
It's done.