Explore the exciting world of AI music generation and how it's changing the way we create music with Meta MusicGen. Using deep learning algorithms, this generative AI music model creates unique and original pieces of music, helping identify and replicate different genres of music.
In this video, Chris Plante explores the exciting world of AI music generation and how it's changing the way we can create music. He dives into Music Gen, an AI music generation system that uses deep learning algorithms to create unique and original pieces of music. The technology behind Music Gen involves feeding a dataset of existing music into an AI model which then learns the patterns and rules of music composition. Chris demonstrates how Music Gen can be found on the Hugging Face tool and generates examples of rock, hip-hop, electronic, jazz, and country music using prompts. He then goes on to replicate the styles of music from famous bands such as Queen, Guns and Roses, The Beatles, and Oasis. The video ends with an outro showcasing Music Gen's ability to extract a broad melody from a reference audio and generate music based on both the description and the provided melody.
welcome to the voice of AI my name is
Chris Plante in this video we explore
the exciting world of AI music
generation and how it's changing the way
that we can create music
we dive into meta's music gen which was
released only this week music gen is an
AI music generation system that uses
deep learning algorithms to create
unique and original pieces of music the
super exciting technology behind music
gen involves feeding a data set of
existing music into an AI model which
then learns the patterns and rules of
music composition this process enables
us to create completely original pieces
of music and helps identify different
genres of music join me as I explore the
future of Music Creation with AI
technology
so today I'm going to give you an
example where music gen can be found on
the hugging face tool a model which will
generate 12 seconds of audio based on
the description that you provide through
your prompt here's a quick intro to show
you how it works
okay so here we are on the
huggingface.co spaces platform and you
can see here this is the music gen model
of by Facebook meta and you can describe
your own music here and you see the
interface and if we look at the first
prompt it's the 80s driving pop star
with heavy drums so if I click on that
you'll see also that the conditional
Melody that's attached to it is
D Minor you'll know it I won't play it
for copyright reasons but let's have a
look and see what it comes out with
[Music]
okay so you've seen what it can do but
now let's try a few test cases and
create a few examples of genres or
styles of music and see what it comes up
with let's try rock hip-hop electronic
Jazz and Country okay so here we are in
chat GPT I thought it would be quite a
good idea to just make sure we get the
right genres of music so first of all
I've asked for five prompts to indicate
individual genres of music so we've been
given them here now you can see one two
three four and five so rock hip-hop
electronic Jazz and Country and now what
I want to do is describe each in a
prompt form that could generate uh
example of each style of music so here
we go we're going to get those prompts
as well so we can simply just copy and
paste the prompts right in to the prompt
generator in music gen so let's take the
first one which is a rock one we'll copy
that out and paste that straight in here
write a song with a catchy guitar riff
powerful drums in the Sweet Child of
Mine Guns and Roses type let's have a
listen
[Music]
okay so that one was pretty cool and it
was quite close I thought to the
original artist
okay now back to chat GPT let's copy out
the hip-hop one and paste that straight
in as well so create a track with a
recurring beat and a set of lyrics with
something like Grandmaster Flash and a
furious five let's have a listen
[Music]
thank you
okay so we're just going to copy the
electronic so let's have a look at this
one use synthesized drums and machines
to create a dance track something like
strobe so let's try it out
[Music]
that was definitely electronic music for
sure okay we'll download those each time
so let's copy the Jazz out this time and
let's uh copy that across so this is by
take five and Dave Brubeck
and let's have a listen to this one
[Music]
okay right that's pretty cool uh very
Jazz like okay let's go back into chat
GPT now we'll copy the country version
number five let's just take that let's
just take that straight back across to
music gen and paste that in and let's
generate that one
[Music]
okay that was very country-esque for
sure and I can definitely see myself
having along to that one right now let's
try recreating some famous stars of
Music I've got some bands in mind
okay so here we are in chat GPT again
and I've given the prompt of you are a
music journalist to chat GPT and I've
asked to describe the music of Queen
because he'll do it much better than I
would do it so that's a pretty good
description and now I'm going to use the
above text to generate a prompt that
will replicate the style of music so you
can see that that's really good so we're
going to just simply copy that and take
it across and paste it into music gen so
here we go describe your music and
straight away will generate let's have a
listen to what this sounds like
[Music]
wow now that is super cool let's listen
again
[Music]
okay that was that was pretty awesome
I've got to say that's the best so far
very anthem-like chord changes Brian
made changes uh typical Roger Taylor
drums so I could imagine that being a
song so let's go back to chat GPT now
describe the music of Guns and Roses the
same again we've just got the
description straight away uh get a good
description and I like to do that so
that we can actually summarize and make
the prompt from that we're going to
summarize that text to make it a bit
thinner
what we'll do now is we'll do the same
problems as well which is use the above
info to generate a prompt so we're
telling chat GPT to write a new song
so let's take that across to music gen
as well now we'll paste that in and
we'll generate
[Music]
wow that was really good too that was a
really good replication of Guns and
Roses
okay so we're going to go back to chat
GPT and now it's described the music of
The Beatles so first of all we'll get
the main characterizations here we go
and once we've got that we can create a
small summary and there we go and we'll
use the above to generate a prompt and
we'll use the above information to
generate the style of The Beatles music
so let's have a look on this one
[Music]
okay so that's kind of interesting it's
got some remnants but I'm not so
convinced so I think what we're going to
try now is we're going to put a shorter
prompt in so let's just half that prompt
and see
[Music]
foreign
[Music]
okay that's even further away I would
say so let's go to the smallest problem
we can get just create a song that is
inspired by The Beatles let's generate
that and see what we get
[Music]
okay cool yep I like that one that is
much more like the styles of The Beatles
so it just shows that the smaller the
prompt sometimes the better the result
okay so back to chat GPT same again
describe the music of Oasis so we do a
little summary and I'm just doing this
to show you how I get to the prompts
because sometimes it's much better to
use chat GPT than to have to even think
about these things so here we go we've
got the Oasis style of music here and
what we're going to do now is we're
going to use the above to generate that
prompt and that will replicate The Oasis
style of music so here we go write a
song with guitar Melodies and obviously
chord progression sing-along choruses
that's exactly what they are so let's
summarize okay so you can see here that
we've got our actual prompt so let's
just copy that out we probably don't
need all of it so we'll just take this
part and we'll copy that across and put
that into
meta's music generation and we'll
generate that off we go let's see what
kind of Oasis track we get here
thank you
[Music]
[Applause]
yeah I think that's pretty good too
Oasis a very good style pretty much
Melodies and chorus progression I would
say that's quite spot on
so finally let's make an outro for this
particular video we can play that at the
end this is where you can optionally
provide a reference audio from which a
broad Melody could be extracted the
model will then try and follow both the
description and the melody provided
let's see how that works okay so here I
am back in music gen so let's put in
Dynamic Cinema style big sound Grand
strings and electronic to create some
really interesting music let's have a
listen to what it's going to sound like
now
[Music]
thank you
Okay so we've saved that file and we're
going to use it as a base for the
condition of the melody for the next one
and generate off the back of that one so
let's see what that one sounds like
[Music]
well that was really good and a great
place to stop well I have to say I've
really enjoyed meta's music gen today
it's been really great and thanks for
hugging face for putting that together
well I guess the debate on AI music will
also continue as musicians Express their
concern on how this will affect them in
the future as well as many of the other
AI models that will affect us all well
that's it for today so thanks very much
for watching and if you haven't already
please like And subscribe to our channel
for more AI related content and tips if
you have any questions or feedback
please leave them in the comments
section below and I'll see you next time
all the very best I'm Chris from the
voice of AI cheers and bye bye now
thank you
Meet AIVA the Artificial Intelligence composing emotional soundtrack music
Eva is an AI software for creating emotional soundtrack music with preset styles, influences, and MIDI file uploads. It offers affordable plans for individuals, students, and enterprises, and includes a gamified composition process and user manuals. Trusted by big players, AIVA is perfect for small and large scale musicians and filmmakers.
The AI Generated Sound Revolution
Advancements in AI audio, particularly the SoundStorm project by Google, offer potential applications in movie dialogues, video games, and digital assistants. Emotional aspect of AI voices is crucial, with companies working to make them sound more human. Adobe's speech enhancement tool eliminates background noise in voice recordings. The audience for AI-related content is mainly male.
Meta’s MMS – Speech2Text & Text2Speech
Meta's MMS project aims to break language barriers by enabling machines to recognize and produce speech in over 1,000 languages. Using Bible translations, it provides labeled data for 1,100 languages and unlabeled data for nearly 4,000 languages, addressing the challenge of finding labeled data. The project's code and models are publicly available for researchers to contribute to preserving language diversity. This project represents a significant step towards achieving the goal of creating a world where technology can understand and communicate in any language, effectively preserving linguistic diversity.
Join our community of happy clients and provide excellent customer support with LiveAgent.
Our website uses cookies. By continuing we assume your permission to deploy cookies as detailed in our privacy and cookies policy.