
yo listen to rap every day but I also
poop we’ve been having any other ways
yeah love computers they’re incredibly
flat let me show you how I build rap
Whitney I took all the music from Iraq
playlists converted into training my
computer for a bit the music that I made
is also that time to get the show on the
road isn’t it my name is Richard Boronia
and I’m a sophomore here at Dorney
Valley High School in my free time I
like to code and go to hackathons but I
also really love listening to music I’ve
published a plethora of independent
Android apps and Alexa skills and most
recently I’ve worked with Stanford
University to predict human linguistic
models using machine learning
now what I’ve noticed about modern music
is that a lot of it sounds really
similar don’t get me wrong there is some
really good music out there but when you
look at the vocals or the background
music the kind of sound was saying it
seems as though we’re approaching a wall
in creating new and unique music which
brings me to my question what will
happen when we run out of innovation you
see innovation is like a marathon and
sooner or later were going to run out of
ideas when that happens artificial
intelligence and machine learning will
provide us with the ideas and
innovations we need of the future when
you do something like send an email or
use your calculator up your computer
takes an input apply some code onto that
input that creates an output now machine
learning is similar but it’s a whole lot
cooler instead of just taking an input
and making an output artificial
intelligence takes a whole lot of input
it takes this training data learns the
patterns in the data and creates a model
based around the patterns that it has
learned now this is cool because it
allows us to create interpretations
based on these models now artificial
intelligence is actually based around
the way humans think similar to how we
as humans use neurons to identify
understand and explain the world around
us
artificial intelligence uses a neural
networks each neuron access an
individual algorithm and the combined
layers of neurons are called neural
networks for example in this neural
network responsible for identifying cats
and dogs
each neural layer narrows down into the
number of classifications here a cat or
a dog now the way it does this is every
neuron produces a completely random
output at first however as we begin to
train the neural network with more and
more data it starts to become smarter
the more training data we provide the
more responsive it becomes to minor
details so I learned about artificial
intelligence recently and I decided I
wanted to teach my computer how to round
or actually I wanted to teach my
computer how to teach itself have around
here’s what I did in the beginning of
winter break I came across a blog post
by a man named Andre Karpov
they outlined how neural networks worked
and it taught me how to install one on
my computer after getting everything set
up I started to move towards my goal
through a series of takes take one
so the neural network that I set up was
meant to take some text input create a
model based around this text and produce
some output based on the model and the
patterns that it learned
the text the fragrant that I had set up
came with an already made vial of
Shakespeare that I could use an input
into my framework so I took it
put it into the machine learning
framework let it train for six hours and
was able to produce this output know if
you’ll read it it sounds like
Shakespeare on drugs but despite the
fact that it doesn’t really make sense
this is entirely computer-generated let
me give you an explanation of what
happened my computer took the
Shakespeare it learned the patterns in
Shakespeare’s writing and created a
model based around the patterns that it
had learned then when I produce some
output they use that model to create
what you see in front of you but this is
kind of boring that if you’ll ask any of
my English teachers they’ll be quick to
tell you that I am NOT a fan of
literature
so forget Shakespeare I wanted to teach
my computer how to rap I thought it
would be easy I would just take my music
from my rap playlist download it turn it
into text put the text into my framework
and get some text output and then make
that some music same way I get the
Shakespeare so after five days of trying
to follow these steps I had done it I
had finally done it
just kidding this is only take two it
turned out I was not able to put my text
input into my sheets into my machine
learning framework because mp3 files
weren’t able to combine as text so I had
to use a different option I completely
wiped out the framework that I was using
and installed a new one one that could
take Wow audio file
as input so I repeated the process I
took all my mp3 files downloaded them
turned them in to walk fast put them
into the framework let it train for six
hours and got some output this time I
was actually able to get something yeah
I was disappointed too it turned out
that since rap music has so much going
on in it like background music vocals
drums and straight fire my artificial
intelligence system was unable to
distinguish the different parts and was
only able to produce an end result of
static maybe if I had more computational
power or more training time I would have
been able to create something that
worked but I didn’t have that
computational strength so I decided to
move on to another try take 4 so this
time I installed a different framework
one that could take many audio files as
input if you’re not sure on what a MIDI
file is it’s basically a file that
contains audio but instead of containing
the actual file of music that contains
sheet using that can be played digitally
through any virtual instrument so I
downloaded hundreds of MIDI files from
rap songs and I put it all into the
artificial engine then I let it train
for nearly 12 hours and produced output
at 10 percent intervals to examine the
progress of my machine learning
framework here’s what a beat sounded
it’s not me it’s done okay this is not a
really really bad but this was only ten
percent and I have continued to let it
train and generate more many files
eventually I was able to get a really
nice tune out of it
and this sounded really nice so I was
pretty proud of it but it’s not really a
rap beat until it has drums so I got my
friend to add some drums to it and mix
it up a little bit here’s what it
sounded like
[Music]
[Applause]
so that sounded a lot better but this
wasn’t actually what I was going for in
the beginning my goal was to create a
rap song this was just a rap
instrumental at this point in the
project I was starting to give up I
thought it wouldn’t be possible and then
I had an idea what if I took the
Shakespeare that I generated earlier and
[Music]
Paul extended why when 42 try upon my
destiny I choose nothing that
just not that much down there being no
other that is being me Thomas Craven
tinniest come back or no test are told
again we must walk we’re saying chase a
lot and work through a very live blood
time to our head if I could guess I
can’t pronounce that as you thought we
made that business away from
that good maid or Meghan and queen
Anderson to compel you movement I joined
a strong boy from TechSoup that’s why
working for more differently as their
arse out tanked that’s just a 40 seconds
in admitted of the full song which I put
on YouTube it’s called computer
generated rap song lil CPU now the end
result was completely different from
what I had planned it took me nearly six
tries to get this going and this was not
what I wanted in the beginning in fact I
wanted a rap song and I didn’t want to
plan out the beat separately from the
lyrics I never planned to use
Shakespeare
that’s my lyrics I never planned to
artificially generate a beat and I
definitely did not plan to spend my
entire winter break on this project
however as I continued to run through
the marathon I was eventually able to
create something that I’m proud of you
see at its current state machine
learning and human intelligence can only
complement to each other however when
you combine artificial intelligence with
natural human curiosity we can innovate
see my computer was able to create some
lyrics and a beat but it required some
human interference to put it all
together into an actual song which
brings about the question what will
happen when we run out of innovation and
what will drive society forward and the
answer is artificial intelligence you
see as AI continues to develop it won’t
push the barriers
innovation and drive society forward in
the marathon of life I truly believe
that just as my rusty old 2011 Mack was
able to create a rap song using machine
learning we as humans will be able to
rely on machine learning for our latest
innovations see it’s just a matter of
time before we start to run out of ideas
and when that happens artificial intelligence has our back thank you
poop we’ve been having any other ways
yeah love computers they’re incredibly
flat let me show you how I build rap
Whitney I took all the music from Iraq
playlists converted into training my
computer for a bit the music that I made
is also that time to get the show on the
road isn’t it my name is Richard Boronia
and I’m a sophomore here at Dorney
Valley High School in my free time I
like to code and go to hackathons but I
also really love listening to music I’ve
published a plethora of independent
Android apps and Alexa skills and most
recently I’ve worked with Stanford
University to predict human linguistic
models using machine learning
now what I’ve noticed about modern music
is that a lot of it sounds really
similar don’t get me wrong there is some
really good music out there but when you
look at the vocals or the background
music the kind of sound was saying it
seems as though we’re approaching a wall
in creating new and unique music which
brings me to my question what will
happen when we run out of innovation you
see innovation is like a marathon and
sooner or later were going to run out of
ideas when that happens artificial
intelligence and machine learning will
provide us with the ideas and
innovations we need of the future when
you do something like send an email or
use your calculator up your computer
takes an input apply some code onto that
input that creates an output now machine
learning is similar but it’s a whole lot
cooler instead of just taking an input
and making an output artificial
intelligence takes a whole lot of input
it takes this training data learns the
patterns in the data and creates a model
based around the patterns that it has
learned now this is cool because it
allows us to create interpretations
based on these models now artificial
intelligence is actually based around
the way humans think similar to how we
as humans use neurons to identify
understand and explain the world around
us
artificial intelligence uses a neural
networks each neuron access an
individual algorithm and the combined
layers of neurons are called neural
networks for example in this neural
network responsible for identifying cats
and dogs
each neural layer narrows down into the
number of classifications here a cat or
a dog now the way it does this is every
neuron produces a completely random
output at first however as we begin to
train the neural network with more and
more data it starts to become smarter
the more training data we provide the
more responsive it becomes to minor
details so I learned about artificial
intelligence recently and I decided I
wanted to teach my computer how to round
or actually I wanted to teach my
computer how to teach itself have around
here’s what I did in the beginning of
winter break I came across a blog post
by a man named Andre Karpov
they outlined how neural networks worked
and it taught me how to install one on
my computer after getting everything set
up I started to move towards my goal
through a series of takes take one
so the neural network that I set up was
meant to take some text input create a
model based around this text and produce
some output based on the model and the
patterns that it learned
the text the fragrant that I had set up
came with an already made vial of
Shakespeare that I could use an input
into my framework so I took it
put it into the machine learning
framework let it train for six hours and
was able to produce this output know if
you’ll read it it sounds like
Shakespeare on drugs but despite the
fact that it doesn’t really make sense
this is entirely computer-generated let
me give you an explanation of what
happened my computer took the
Shakespeare it learned the patterns in
Shakespeare’s writing and created a
model based around the patterns that it
had learned then when I produce some
output they use that model to create
what you see in front of you but this is
kind of boring that if you’ll ask any of
my English teachers they’ll be quick to
tell you that I am NOT a fan of
literature
so forget Shakespeare I wanted to teach
my computer how to rap I thought it
would be easy I would just take my music
from my rap playlist download it turn it
into text put the text into my framework
and get some text output and then make
that some music same way I get the
Shakespeare so after five days of trying
to follow these steps I had done it I
had finally done it
just kidding this is only take two it
turned out I was not able to put my text
input into my sheets into my machine
learning framework because mp3 files
weren’t able to combine as text so I had
to use a different option I completely
wiped out the framework that I was using
and installed a new one one that could
take Wow audio file
as input so I repeated the process I
took all my mp3 files downloaded them
turned them in to walk fast put them
into the framework let it train for six
hours and got some output this time I
was actually able to get something yeah
I was disappointed too it turned out
that since rap music has so much going
on in it like background music vocals
drums and straight fire my artificial
intelligence system was unable to
distinguish the different parts and was
only able to produce an end result of
static maybe if I had more computational
power or more training time I would have
been able to create something that
worked but I didn’t have that
computational strength so I decided to
move on to another try take 4 so this
time I installed a different framework
one that could take many audio files as
input if you’re not sure on what a MIDI
file is it’s basically a file that
contains audio but instead of containing
the actual file of music that contains
sheet using that can be played digitally
through any virtual instrument so I
downloaded hundreds of MIDI files from
rap songs and I put it all into the
artificial engine then I let it train
for nearly 12 hours and produced output
at 10 percent intervals to examine the
progress of my machine learning
framework here’s what a beat sounded
it’s not me it’s done okay this is not a
really really bad but this was only ten
percent and I have continued to let it
train and generate more many files
eventually I was able to get a really
nice tune out of it
and this sounded really nice so I was
pretty proud of it but it’s not really a
rap beat until it has drums so I got my
friend to add some drums to it and mix
it up a little bit here’s what it
sounded like
[Music]
[Applause]
so that sounded a lot better but this
wasn’t actually what I was going for in
the beginning my goal was to create a
rap song this was just a rap
instrumental at this point in the
project I was starting to give up I
thought it wouldn’t be possible and then
I had an idea what if I took the
Shakespeare that I generated earlier and
[Music]
Paul extended why when 42 try upon my
destiny I choose nothing that
just not that much down there being no
other that is being me Thomas Craven
tinniest come back or no test are told
again we must walk we’re saying chase a
lot and work through a very live blood
time to our head if I could guess I
can’t pronounce that as you thought we
made that business away from
that good maid or Meghan and queen
Anderson to compel you movement I joined
a strong boy from TechSoup that’s why
working for more differently as their
arse out tanked that’s just a 40 seconds
in admitted of the full song which I put
on YouTube it’s called computer
generated rap song lil CPU now the end
result was completely different from
what I had planned it took me nearly six
tries to get this going and this was not
what I wanted in the beginning in fact I
wanted a rap song and I didn’t want to
plan out the beat separately from the
lyrics I never planned to use
Shakespeare
that’s my lyrics I never planned to
artificially generate a beat and I
definitely did not plan to spend my
entire winter break on this project
however as I continued to run through
the marathon I was eventually able to
create something that I’m proud of you
see at its current state machine
learning and human intelligence can only
complement to each other however when
you combine artificial intelligence with
natural human curiosity we can innovate
see my computer was able to create some
lyrics and a beat but it required some
human interference to put it all
together into an actual song which
brings about the question what will
happen when we run out of innovation and
what will drive society forward and the
answer is artificial intelligence you
see as AI continues to develop it won’t
push the barriers
innovation and drive society forward in
the marathon of life I truly believe
that just as my rusty old 2011 Mack was
able to create a rap song using machine
learning we as humans will be able to
rely on machine learning for our latest
innovations see it’s just a matter of
time before we start to run out of ideas
and when that happens artificial intelligence has our back thank you
Please follow and like us: