hi my name is Curtis Hawthorne I work at Google brain on the magenta project with
Google brain on the magenta project with
Google brain on the magenta project with magenta project we’re seeking to answer
magenta project we’re seeking to answer
magenta project we’re seeking to answer the questions can machine learning be
the questions can machine learning be
the questions can machine learning be used to create compelling pieces of
used to create compelling pieces of
used to create compelling pieces of music and art if so how and if not why
music and art if so how and if not why
music and art if so how and if not why not now these are easy questions to ask
not now these are easy questions to ask
not now these are easy questions to ask but surprisingly difficult to answer a
but surprisingly difficult to answer a
but surprisingly difficult to answer a truly compelling piece of music or art
truly compelling piece of music or art
truly compelling piece of music or art transcends a generated data point music
transcends a generated data point music
transcends a generated data point music and art connects us to one another
and art connects us to one another
and art connects us to one another engages us surprises us and provides a
engages us surprises us and provides a
engages us surprises us and provides a means of expression for some of our most
means of expression for some of our most
means of expression for some of our most personal and powerful stories can a
personal and powerful stories can a
personal and powerful stories can a machine learn to do that well it turns
machine learn to do that well it turns
machine learn to do that well it turns out we aren’t the first people to ask
out we aren’t the first people to ask
out we aren’t the first people to ask these questions people have been trying
these questions people have been trying
these questions people have been trying to compose music using algorithms for a
to compose music using algorithms for a
to compose music using algorithms for a long time back in the 1700s musical dice
long time back in the 1700s musical dice
long time back in the 1700s musical dice games were very popular
games were very popular
games were very popular they allowed any player to compose a
they allowed any player to compose a
they allowed any player to compose a piece of music just by rolling some dice
piece of music just by rolling some dice
piece of music just by rolling some dice now of course the compositions that
now of course the compositions that
now of course the compositions that these games produced were limited to the
these games produced were limited to the
these games produced were limited to the snippets of music written by the creator
snippets of music written by the creator
snippets of music written by the creator of the game so in the end most of the
of the game so in the end most of the
of the game so in the end most of the composition sounded fairly similar to
composition sounded fairly similar to
composition sounded fairly similar to one another one way to solve this
one another one way to solve this
one another one way to solve this problem of limited compositions is to
problem of limited compositions is to
problem of limited compositions is to make the rules for producing the
make the rules for producing the
make the rules for producing the compositions much more complex as far
compositions much more complex as far
compositions much more complex as far back as the 1800s Ada Lovelace was
back as the 1800s Ada Lovelace was
back as the 1800s Ada Lovelace was writing down ideas about how computers
writing down ideas about how computers
writing down ideas about how computers could one day be used to create
could one day be used to create
could one day be used to create elaborate and scientific pieces of music
elaborate and scientific pieces of music
elaborate and scientific pieces of music in the 1950s researchers at the
in the 1950s researchers at the
in the 1950s researchers at the University of Illinois I finally got a
University of Illinois I finally got a
University of Illinois I finally got a chance to try out some of those ideas
chance to try out some of those ideas
chance to try out some of those ideas they use there iliacs supercomputer to
they use there iliacs supercomputer to
they use there iliacs supercomputer to write music for a string quartet and the
write music for a string quartet and the
write music for a string quartet and the music that they produced was much more
music that they produced was much more
music that they produced was much more interesting than complex than dice game
interesting than complex than dice game
interesting than complex than dice game music and researchers and musicians have
music and researchers and musicians have
music and researchers and musicians have been exploring those ideas ever since
been exploring those ideas ever since
been exploring those ideas ever since however this approach still has a
however this approach still has a
however this approach still has a fundamental limitation the computer can
fundamental limitation the computer can
fundamental limitation the computer can only write music that follows the rules
only write music that follows the rules
only write music that follows the rules of the algorithm programmed into it so
of the algorithm programmed into it so
of the algorithm programmed into it so if you’ve carefully written an algorithm
if you’ve carefully written an algorithm
if you’ve carefully written an algorithm for producing string quartet music that
for producing string quartet music that
for producing string quartet music that same algorithm isn’t going to wake up
same algorithm isn’t going to wake up
same algorithm isn’t going to wake up one day and decide to write some dubstep
one day and decide to write some dubstep
one day and decide to write some dubstep what if there was a way to have
what if there was a way to have
what if there was a way to have computers write music the same way that
computers write music the same way that
computers write music the same way that humans write music so instead of using
humans write music so instead of using
humans write music so instead of using handcrafted algorithms and the way that
handcrafted algorithms and the way that
handcrafted algorithms and the way that humans write music is first we listen to
humans write music is first we listen to
humans write music is first we listen to music thousands and thousands of hours
music thousands and thousands of hours
music thousands and thousands of hours of music from the earliest stages of our
of music from the earliest stages of our
of music from the earliest stages of our lives
lives
lives subconsciously picking up on patterns
subconsciously picking up on patterns
subconsciously picking up on patterns and learning about different genres then
and learning about different genres then
and learning about different genres then when we want to write something new
when we want to write something new
when we want to write something new we can draw on that experience and our
we can draw on that experience and our
we can draw on that experience and our own inspiration to write a new
own inspiration to write a new
own inspiration to write a new composition this idea of having
composition this idea of having
composition this idea of having computers learn not from handcrafted
computers learn not from handcrafted
computers learn not from handcrafted algorithms but from data is called
algorithms but from data is called
algorithms but from data is called machine learning
machine learning
machine learning but how can a machine learn from data
but how can a machine learn from data
but how can a machine learn from data there are many ways that this can be
there are many ways that this can be
there are many ways that this can be done but let me give you one quick
done but let me give you one quick
done but let me give you one quick example let’s say that you wanted to
example let’s say that you wanted to
example let’s say that you wanted to write a system that could tell you
write a system that could tell you
write a system that could tell you whether a given photo was of a cat or of
whether a given photo was of a cat or of
whether a given photo was of a cat or of an elephant and let’s say that you
an elephant and let’s say that you
an elephant and let’s say that you already had systems in place that could
already had systems in place that could
already had systems in place that could recognize features such as fertilize if
recognize features such as fertilize if
recognize features such as fertilize if you’ve had enough photos of elephants
you’ve had enough photos of elephants
you’ve had enough photos of elephants into this system only telling the system
into this system only telling the system
into this system only telling the system that these are photos of elephants over
that these are photos of elephants over
that these are photos of elephants over time the system would learn that
time the system would learn that
time the system would learn that features like tusks and eyes are highly
features like tusks and eyes are highly
features like tusks and eyes are highly correlated with an elephant being in the
correlated with an elephant being in the
correlated with an elephant being in the photo similarly if you fit in enough
photo similarly if you fit in enough
photo similarly if you fit in enough photos of cats over time the system
photos of cats over time the system
photos of cats over time the system would learn that fur and eyes are highly
would learn that fur and eyes are highly
would learn that fur and eyes are highly correlated with a cat being in the photo
correlated with a cat being in the photo
correlated with a cat being in the photo now in this system the the important
now in this system the the important
now in this system the the important point is that we didn’t write an
point is that we didn’t write an
point is that we didn’t write an algorithm that listed out the specific
algorithm that listed out the specific
algorithm that listed out the specific features for each animal instead the
features for each animal instead the
features for each animal instead the system learned on its own which features
system learned on its own which features
system learned on its own which features correspond to which animal and it even
correspond to which animal and it even
correspond to which animal and it even learned that some features in this case
learned that some features in this case
learned that some features in this case eyes correspond to both animals now in
eyes correspond to both animals now in
eyes correspond to both animals now in this example we have just a few of these
this example we have just a few of these
this example we have just a few of these kinds of nodes but if you take many of
kinds of nodes but if you take many of
kinds of nodes but if you take many of these kinds of nodes sometimes many
these kinds of nodes sometimes many
these kinds of nodes sometimes many thousands of these nodes and arrange
thousands of these nodes and arrange
thousands of these nodes and arrange them in multiple layers you create
them in multiple layers you create
them in multiple layers you create what’s sometimes called an artificial
what’s sometimes called an artificial
what’s sometimes called an artificial neural network now it’s called this
neural network now it’s called this
neural network now it’s called this because it’s loosely based on the way
because it’s loosely based on the way
because it’s loosely based on the way that biological neurons work together in
that biological neurons work together in
that biological neurons work together in the human brain to solve problems now
the human brain to solve problems now
the human brain to solve problems now the system that we have here has has
the system that we have here has has
the system that we have here has has many flaws there are many situations
many flaws there are many situations
many flaws there are many situations where it won’t work for example not all
where it won’t work for example not all
where it won’t work for example not all cats have fur and I’m sure you could
cats have fur and I’m sure you could
cats have fur and I’m sure you could come up with a long list of other
come up with a long list of other
come up with a long list of other examples that’s why for real image
examples that’s why for real image
examples that’s why for real image recognition systems we take those
recognition systems we take those
recognition systems we take those neurons that I was talking about we
neurons that I was talking about we
neurons that I was talking about we arrange them in multiple layers and then
arrange them in multiple layers and then
arrange them in multiple layers and then we let the system learn on its own which
we let the system learn on its own which
we let the system learn on its own which features are important for recognizing
features are important for recognizing
features are important for recognizing objects and photos usually the first few
objects and photos usually the first few
objects and photos usually the first few layers learn to recognize things like
layers learn to recognize things like
layers learn to recognize things like lines and basic shapes then the
lines and basic shapes then the
lines and basic shapes then the intermediate layers are able to piece
intermediate layers are able to piece
intermediate layers are able to piece together those lines in basic shapes and
together those lines in basic shapes and
together those lines in basic shapes and recognize more complex shapes and
recognize more complex shapes and
recognize more complex shapes and patterns and by the time that we get to
patterns and by the time that we get to
patterns and by the time that we get to the final few layers it’s able to detect
the final few layers it’s able to detect
the final few layers it’s able to detect things like spirals there’s something
things like spirals there’s something
things like spirals there’s something there that looks kind of like a bird
there that looks kind of like a bird
there that looks kind of like a bird maybe some honeycomb in the lower left
maybe some honeycomb in the lower left
maybe some honeycomb in the lower left all right so now we have a way for
all right so now we have a way for
all right so now we have a way for machines to learn based on data but how
machines to learn based on data but how
machines to learn based on data but how can we apply that machine learning to
can we apply that machine learning to
can we apply that machine learning to creation of art here’s an example of art
creation of art here’s an example of art
creation of art here’s an example of art created by another team at Google they
created by another team at Google they
created by another team at Google they call their system deep dream it works by
call their system deep dream it works by
call their system deep dream it works by taking a neural network that’s trained
taking a neural network that’s trained
taking a neural network that’s trained to recognize objects and faces and it
to recognize objects and faces and it
to recognize objects and faces and it runs that network on an image in such a
runs that network on an image in such a
runs that network on an image in such a way that it dreams up objects and faces
way that it dreams up objects and faces
way that it dreams up objects and faces that aren’t actually there and in fact
that aren’t actually there and in fact
that aren’t actually there and in fact there was a recent charity art show in
there was a recent charity art show in
there was a recent charity art show in San Francisco where people paid
San Francisco where people paid
San Francisco where people paid thousands of dollars to take home deep
thousands of dollars to take home deep
thousands of dollars to take home deep dream generated artwork so clearly we’re
dream generated artwork so clearly we’re
dream generated artwork so clearly we’re at a point where machine generated art
at a point where machine generated art
at a point where machine generated art is able to make some kind of human
is able to make some kind of human
is able to make some kind of human artistic connection all right now how do
artistic connection all right now how do
artistic connection all right now how do we apply machine learning and neural
we apply machine learning and neural
we apply machine learning and neural networks to music creation one way to
networks to music creation one way to
networks to music creation one way to approach this problem is as a prediction
approach this problem is as a prediction
approach this problem is as a prediction problem so you can say given a sequence
problem so you can say given a sequence
problem so you can say given a sequence of notes predict what the most likely
of notes predict what the most likely
of notes predict what the most likely next note is in order to do this we’ll
next note is in order to do this we’ll
next note is in order to do this we’ll need to use a slightly different kind of
need to use a slightly different kind of
need to use a slightly different kind of neural network than the one I’ve been
neural network than the one I’ve been
neural network than the one I’ve been talking about so far we’ll need to use
talking about so far we’ll need to use
talking about so far we’ll need to use what’s called a recurrent neural network
what’s called a recurrent neural network
what’s called a recurrent neural network and this type of network has access to
and this type of network has access to
and this type of network has access to not only the information that it’s
not only the information that it’s
not only the information that it’s currently being presented with so in
currently being presented with so in
currently being presented with so in this case the most recent note of the
this case the most recent note of the
this case the most recent note of the melody but it also has access to a
melody but it also has access to a
melody but it also has access to a representation of information in the
representation of information in the
representation of information in the past so the previous notes of the melody
past so the previous notes of the melody
past so the previous notes of the melody with this type of network you can ask
with this type of network you can ask
with this type of network you can ask the question given a sequence of notes
the question given a sequence of notes
the question given a sequence of notes what’s the next note and then if you
what’s the next note and then if you
what’s the next note and then if you repeat that process and ask it to keep
repeat that process and ask it to keep
repeat that process and ask it to keep predicting the next note you can
predicting the next note you can
predicting the next note you can generate a whole new melody all right so
generate a whole new melody all right so
generate a whole new melody all right so now we’ve got a way to train a machine
now we’ve got a way to train a machine
now we’ve got a way to train a machine to produce a melody we just need some
to produce a melody we just need some
to produce a melody we just need some training data so we collected a library
training data so we collected a library
training data so we collected a library of songs tens of thousands of songs
of songs tens of thousands of songs
of songs tens of thousands of songs everything from Bach to Radiohead and we
everything from Bach to Radiohead and we
everything from Bach to Radiohead and we took the melody lines from those songs
took the melody lines from those songs
took the melody lines from those songs and we fed them into the network one
and we fed them into the network one
and we fed them into the network one note at a time and then we as a as a
note at a time and then we as a as a
note at a time and then we as a as a starting point we gave the network the
starting point we gave the network the
starting point we gave the network the first four notes of twinkle twinkle
first four notes of twinkle twinkle
first four notes of twinkle twinkle little star dun dun dun dun and asked it
little star dun dun dun dun and asked it
little star dun dun dun dun and asked it to continue the melody and this is what
to continue the melody and this is what
to continue the melody and this is what it came up with
[Music]
[Music] all right go to our website you can
all right go to our website you can
all right go to our website you can listen to the rest of it there all right
listen to the rest of it there all right
listen to the rest of it there all right so that’s pretty good but we wanted to
so that’s pretty good but we wanted to
so that’s pretty good but we wanted to explore ways that we could make melody
explore ways that we could make melody
explore ways that we could make melody generation even better focusing on just
generation even better focusing on just
generation even better focusing on just next step note prediction has a few
next step note prediction has a few
next step note prediction has a few important limitations one is long term
important limitations one is long term
important limitations one is long term structure if you’re only asking the
structure if you’re only asking the
structure if you’re only asking the model to predict the next note it has
model to predict the next note it has
model to predict the next note it has difficulty coming up with things that we
difficulty coming up with things that we
difficulty coming up with things that we like to listen for in music like
like to listen for in music like
like to listen for in music like repeated motifs and longer melodic arcs
repeated motifs and longer melodic arcs
repeated motifs and longer melodic arcs another problem is repetition if your
another problem is repetition if your
another problem is repetition if your training data has a lot of repeated
training data has a lot of repeated
training data has a lot of repeated notes and you can hear this for example
notes and you can hear this for example
notes and you can hear this for example even in twinkle twinkle little star dun
even in twinkle twinkle little star dun
even in twinkle twinkle little star dun dun dun dun then over time the model
dun dun dun then over time the model
dun dun dun then over time the model might learn that given a particular note
might learn that given a particular note
might learn that given a particular note if you ask it to predict the next note a
if you ask it to predict the next note a
if you ask it to predict the next note a reasonable prediction might be well just
reasonable prediction might be well just
reasonable prediction might be well just repeat that same note similarly if the
repeat that same note similarly if the
repeat that same note similarly if the last thing that happened was arrest or
last thing that happened was arrest or
last thing that happened was arrest or silence a reasonable prediction might be
silence a reasonable prediction might be
silence a reasonable prediction might be well just add another rest extend the
well just add another rest extend the
well just add another rest extend the silence and this works well in the
silence and this works well in the
silence and this works well in the short-term but as you can imagine it
short-term but as you can imagine it
short-term but as you can imagine it breaks down in the long-term so one of
breaks down in the long-term so one of
breaks down in the long-term so one of our interns this last summer Natasha
our interns this last summer Natasha
our interns this last summer Natasha Jake’s came up with the idea of using a
Jake’s came up with the idea of using a
Jake’s came up with the idea of using a technique called reinforcement learning
technique called reinforcement learning
technique called reinforcement learning to improve our models now reinforcement
to improve our models now reinforcement
to improve our models now reinforcement learning works a little bit different
learning works a little bit different
learning works a little bit different than the training method I’ve been
than the training method I’ve been
than the training method I’ve been talking about so far instead of showing
talking about so far instead of showing
talking about so far instead of showing the model a list of examples and saying
the model a list of examples and saying
the model a list of examples and saying these examples are correct instead we
these examples are correct instead we
these examples are correct instead we ask the model to produce some output and
ask the model to produce some output and
ask the model to produce some output and then we give it a reward mathematically
then we give it a reward mathematically
then we give it a reward mathematically speaking based on how well it did this
speaking based on how well it did this
speaking based on how well it did this is conceptually similar to the idea the
is conceptually similar to the idea the
is conceptually similar to the idea the composition student and their teacher
composition student and their teacher
composition student and their teacher the student writes composition and then
the student writes composition and then
the student writes composition and then the teacher gives them feedback on how
the teacher gives them feedback on how
the teacher gives them feedback on how good or bad the composition is so using
good or bad the composition is so using
good or bad the composition is so using this system over time the model will
this system over time the model will
this system over time the model will learn to favor producing outputs that
learn to favor producing outputs that
learn to favor producing outputs that receive a positive reward now of course
receive a positive reward now of course
receive a positive reward now of course the important part of this is we have to
the important part of this is we have to
the important part of this is we have to define what sounds good is we certainly
define what sounds good is we certainly
define what sounds good is we certainly didn’t have time for a human to sit
didn’t have time for a human to sit
didn’t have time for a human to sit around and listen to all of the
around and listen to all of the
around and listen to all of the thousands upon thousands of melodies
thousands upon thousands of melodies
thousands upon thousands of melodies that the model produces and rate them so
that the model produces and rate them so
that the model produces and rate them so what we did is we found a composition
what we did is we found a composition
what we did is we found a composition textbook and we just took the list of it
textbook and we just took the list of it
textbook and we just took the list of it listed some rules for what makes a good
listed some rules for what makes a good
listed some rules for what makes a good melody and we took that list and
melody and we took that list and
melody and we took that list and programmed it in and of course the first
programmed it in and of course the first
programmed it in and of course the first rule that it lists is avoid excessive
rule that it lists is avoid excessive
rule that it lists is avoid excessive repeated notes just one of the problems
repeated notes just one of the problems
repeated notes just one of the problems we were trying to solve ok so let’s see
we were trying to solve ok so let’s see
we were trying to solve ok so let’s see how well we did first I’ll play you an
how well we did first I’ll play you an
how well we did first I’ll play you an example of what the model produced
example of what the model produced
example of what the model produced before reinforcement learning
before reinforcement learning
before reinforcement learning you’ll notice that the melody isn’t very
you’ll notice that the melody isn’t very
you’ll notice that the melody isn’t very end
end
end and there’s a long silence in the middle
all right and here’s what it sounds like
all right and here’s what it sounds like after reinforcement learning all right a
after reinforcement learning all right a
after reinforcement learning all right a lot more interesting after we posted
lot more interesting after we posted
lot more interesting after we posted that result online MIT Technology Review
that result online MIT Technology Review
that result online MIT Technology Review said that that particular melody sounded
said that that particular melody sounded
said that that particular melody sounded a bit like a jingle from the latest
a bit like a jingle from the latest
a bit like a jingle from the latest toothpaste campaign which we take as a
toothpaste campaign which we take as a
toothpaste campaign which we take as a high compliment all right so that’s far
high compliment all right so that’s far
high compliment all right so that’s far I’ve been talking about music but
I’ve been talking about music but
I’ve been talking about music but magenta is interested in visual art as
magenta is interested in visual art as
magenta is interested in visual art as well
well
well a recent breakthrough in machine
a recent breakthrough in machine
a recent breakthrough in machine learning has been the ability to
learning has been the ability to
learning has been the ability to transfer the style of one image to
transfer the style of one image to
transfer the style of one image to another so for example you can see what
another so for example you can see what
another so for example you can see what any given photo looks like painted in
any given photo looks like painted in
any given photo looks like painted in the style of starry night we wanted to
the style of starry night we wanted to
the style of starry night we wanted to take this idea a little bit further and
take this idea a little bit further and
take this idea a little bit further and so in collaboration with another team at
so in collaboration with another team at
so in collaboration with another team at Google brain we developed a network that
Google brain we developed a network that
Google brain we developed a network that could learn multiple styles and mix them
could learn multiple styles and mix them
could learn multiple styles and mix them together this is a video of peekaboo
together this is a video of peekaboo
together this is a video of peekaboo my teammates dog transformed to look
my teammates dog transformed to look
my teammates dog transformed to look like several different painting styles
like several different painting styles
like several different painting styles so you’ll see the the slider for starry
so you’ll see the the slider for starry
so you’ll see the the slider for starry night just got turned up to 100% it’s
night just got turned up to 100% it’s
night just got turned up to 100% it’s being mixed with a few other painting
being mixed with a few other painting
being mixed with a few other painting styles and then we’ll remove the starry
styles and then we’ll remove the starry
styles and then we’ll remove the starry night style from the transformed image
night style from the transformed image
night style from the transformed image all right I can go to the next slide
all right I can go to the next slide
all right I can go to the next slide okay and we’re working on a bunch of
okay and we’re working on a bunch of
okay and we’re working on a bunch of other projects too but I don’t have time
other projects too but I don’t have time
other projects too but I don’t have time to talk about all of them today so what
to talk about all of them today so what
to talk about all of them today so what we’ve got here is pretty good but
we’ve got here is pretty good but
we’ve got here is pretty good but remember what I talked about at the
remember what I talked about at the
remember what I talked about at the beginning about how music and art
beginning about how music and art
beginning about how music and art connects us to one another engages us
connects us to one another engages us
connects us to one another engages us surprises us has what we’ve done here
surprises us has what we’ve done here
surprises us has what we’ve done here accomplished that perhaps not on its own
accomplished that perhaps not on its own
accomplished that perhaps not on its own but we believe the key to magentas
but we believe the key to magentas
but we believe the key to magentas success will be the community we
success will be the community we
success will be the community we recognize that music and art is too big
recognize that music and art is too big
recognize that music and art is too big of a problem to tackle on our own so
of a problem to tackle on our own so
of a problem to tackle on our own so we’re building a community of musicians
we’re building a community of musicians
we’re building a community of musicians coders machine learning researchers to
coders machine learning researchers to
coders machine learning researchers to work on this together we’ve posted the
work on this together we’ve posted the
work on this together we’ve posted the code for our models up on github and we
code for our models up on github and we
code for our models up on github and we encourage anyone to try it out tweak it
encourage anyone to try it out tweak it
encourage anyone to try it out tweak it use it improve it in ways that maybe we
use it improve it in ways that maybe we
use it improve it in ways that maybe we wouldn’t have thought of here’s a great
wouldn’t have thought of here’s a great
wouldn’t have thought of here’s a great example of how community can take a
example of how community can take a
example of how community can take a relatively simple beginning and
relatively simple beginning and
relatively simple beginning and transform it into something really
transform it into something really
transform it into something really interesting
interesting
interesting you remember that tune from earlier done
you remember that tune from earlier done
you remember that tune from earlier done dunh dunh well when we posted that
dunh dunh well when we posted that
dunh dunh well when we posted that example online Jason Wynn who studies
example online Jason Wynn who studies
example online Jason Wynn who studies Vietnamese traditional music wanted to
Vietnamese traditional music wanted to
Vietnamese traditional music wanted to play along with it so he and another
play along with it so he and another
play along with it so he and another artist collaborated and created this
artist collaborated and created this
artist collaborated and created this video
video
video [Music]
all right that’s great we love how the
all right that’s great we love how the tune the magenta created inspired these
tune the magenta created inspired these
tune the magenta created inspired these musicians to come together and create
musicians to come together and create
musicians to come together and create something new next slide please
something new next slide please
something new next slide please and we wanted to expand on this idea of
and we wanted to expand on this idea of
and we wanted to expand on this idea of musicians interacting with magenta so
musicians interacting with magenta so
musicians interacting with magenta so for the neural information processing
for the neural information processing
for the neural information processing systems conference this past December we
systems conference this past December we
systems conference this past December we put together a demo that allows a
put together a demo that allows a
put together a demo that allows a musician to play with magenta in
musician to play with magenta in
musician to play with magenta in real-time so in this video and the
real-time so in this video and the
real-time so in this video and the electric piano sound that you hear will
electric piano sound that you hear will
electric piano sound that you hear will be the human performer plain and melody
be the human performer plain and melody
be the human performer plain and melody and then magenta will respond in
and then magenta will respond in
and then magenta will respond in real-time and continue that melody
alright next like this we think that
alright next like this we think that these kinds of interactions are where
these kinds of interactions are where
these kinds of interactions are where things get really interesting with
things get really interesting with
things get really interesting with magenta we think that machine learning
magenta we think that machine learning
magenta we think that machine learning applied to music and art is going to be
applied to music and art is going to be
applied to music and art is going to be similar to win Rickenbacker and Gibson
similar to win Rickenbacker and Gibson
similar to win Rickenbacker and Gibson figured out the electric guitar they
figured out the electric guitar they
figured out the electric guitar they created this new musical instrument but
created this new musical instrument but
created this new musical instrument but really all they were trying to do was
really all they were trying to do was
really all they were trying to do was come up with a way to make the acoustic
come up with a way to make the acoustic
come up with a way to make the acoustic guitar sound a little bit louder they
guitar sound a little bit louder they
guitar sound a little bit louder they could have never imagined what artists
could have never imagined what artists
could have never imagined what artists like Jimi Hendrix or Saint Vincent would
like Jimi Hendrix or Saint Vincent would
like Jimi Hendrix or Saint Vincent would end up doing with it
end up doing with it
end up doing with it similarly we think that the tools that
similarly we think that the tools that
similarly we think that the tools that machine learning researchers will come
machine learning researchers will come
machine learning researchers will come up with for working with music and art
up with for working with music and art
up with for working with music and art will be used by artists in ways that
will be used by artists in ways that
will be used by artists in ways that those researchers could have never
those researchers could have never
those researchers could have never expected and I hope that some of those
expected and I hope that some of those
expected and I hope that some of those artists are right here today everything
artists are right here today everything
artists are right here today everything that I’ve talked about is up on our
that I’ve talked about is up on our
that I’ve talked about is up on our website the code is all up on github you
website the code is all up on github you
website the code is all up on github you can download all of these experiments
can download all of these experiments
can download all of these experiments and run them on your own computer see
and run them on your own computer see
and run them on your own computer see for yourself how they work try them out
for yourself how they work try them out
for yourself how they work try them out maybe you’ve got a great idea for how
maybe you’ve got a great idea for how
maybe you’ve got a great idea for how much into could be used to create new
much into could be used to create new
much into could be used to create new music and art go ahead try it out I’m
music and art go ahead try it out I’m
music and art go ahead try it out I’m looking forward to seeing what you
looking forward to seeing what you
looking forward to seeing what you create thank you
”
Be First to Comment