I’ve spent the last couple of years actively watching the developments in the world of blockchain and how this spills over into the world of music.
This article takes a deep dive into one of the opportunities that I’ve noticed arising as a result of the shifts that the industry is seeing today – and that’s generative music NFTs.
It seems that web3 is poised to create massive changes in the music industry, and if the massive interest in generative art NFTs seen over the past year is anything to go by, generative music NFTs will definitely be an area to keep an eye on.
In this article I’ll lay out the concept of generative music and how you as an artist can play around with the ideas. I plan on writing more about the concept as I get more involved, as well as sharing my thoughts on other web3 music topics, so if that sounds of interest make sure you sign up to the newsletter and follow along on Twitter.
What is Generative Music?
Generative music is not a new concept. It seems that the term was actually coined by Brian Eno in 1995 to describe ‘any music that is ever-different and changing, created by a system’.
The term tends to be used more broadly today, and generally refers to music that involves a system, generally some set of rules to create variations of a sound or song.
While the term has been around for a while, it’s never been used too much – in fact looking at Google trends we can see a downwards trend in interest since 2004 (the earliest date range available), apart from a slight spike in October 2018.
Note that the chart shows relative interest levels, with 100 being the highest point of interest during that time range.
The renewed interest in generative music that I see online today seems to be a byproduct of the increased focus on generative art – a term that has seen an increased usage as it ties in very much with digital art and NFTs.
If we look at the same period on Google trends for ‘generative art’ we can see this trend:
The massive increase in growth coincides with the NFT boom of 2021 and it’s not a huge leap to expect the same trend to happen over the next year or so when it comes to music.
There are a few reasons why the music side is trailing behind art – it’s arguably more complex from a technical viewpoint, but the scene seems to be catching up fairly quickly – and given the cultural significance that music has in the world I expect adoption to increase significantly.
Generative music is going to become a valuable concept for web3 musicians to get to grips with. Later on in this post I will go into more detail about how to create it, which should hopefully be helpful to artists (or anyone) trying to familiarise themselves with the process.
What are NFTs?
For those of you reading this who aren’t familiar with NFTs, here’s the summary from Wikipedia:
“A non-fungible token (NFT) is a non-interchangeable unit of data stored on a blockchain, a form of digital ledger, that can be sold and traded. Types of NFT data units may be associated with digital files such as photos, videos, and audio.”
To sum it up in an easier way, if cryptocurrency is the digital equivalent of the money you own offline, NFTs are the digital equivalent of all the other things you own offline – the house you own, the art you hang up, the music you listen to, the clothes you wear.
As we gradually move towards the concept of a metaverse, digital goods are going to become more prevalent. Whether you like it or not, it’s hard to argue with the concept. All of us collect things of meaning in the real world, and as the digital world slowly accounts for more of our lives, why shouldn’t that carry over?
How Does this Apply to Music?
Following on from the example above, you may be wondering how this applies to music.
After all, music for the most part is already digitized, with most people paying a small monthly subscription to access as much as they want – and some of the more passionate fans investing in merchandise and live shows.
I would argue that the small investment currently made by people isn’t an accurate valuation of the true value of music to society, but rather a product of the technological landscape that has evolved over the last 20 years.
NFTs open up the door to a load of new opportunities for fan engagement.
Some examples of this are the option to invest in your favorite artists and share in their success through royalties or community tokens, buy NFTs that grant you access to IRL experiences, or even own exclusive tracks from artists that no one else has heard.
All of these things were arguably conceivable before the advent of web3 (think Bowie Bonds, crowdfunding campaigns or VIP meet and greets) – but it’s the development of this new technology that adds utility to these things and makes them far more engaging, with the immutability of the blockchain to help verify ownership and facilitate transfer of it further down the line.
This new technology adds utility across applications, allowing you to take your digital tokens and move them with you across whatever platforms start to emerge.
These things all become part of your online identity, and the tech allows you to carry these things throughout the different platforms you engage with.
Don’t get me wrong, streaming services will continue to exist, and for most people they will probably be as far as their investment in music goes.
But for those fans who engage enough to invest in merchandise, live shows, and collectables, web3 opens up a whole new way of interacting with, and supporting the artists they love.
Why Generative Music NFTs Have Potential
Music NFTs are already showing that they can be a way for artists to earn a living from their craft, but there is still a way to go both in terms of encouraging user adoption, and also creating a compelling enough offer that it manages to encourage non web3 natives to get involved.
One of the current challenges I see with music NFTs is rarity – and how there is a fine balance between the quality of the artwork and the price. If you want to own a 1/1 NFT of a song, the cost is generally a lot higher because the artist only has that one unique version of the track to sell.
If we take cryptopunks or BAYC, or any of the most successful generative art NFT projects, something they all have in common is that each artwork is unique.
When you buy a cryptopunk, you, and you alone own that punk. The market may set the floor price, but ultimately you decide how much that is worth to you if you wish to sell it. No one can come along with the same punk and offer a different price.
However with music NFTs, a lot of projects offer the opportunity to buy one of a certain number of the same audio.
If we take a look at The Weeknd x Strangeloop Studios collection on Opensea, what you’re actually buying is 1 of 423, 500 or even 1749 versions of the same audio/visual pairing, assigned to its own unique token.
That means that there will still be a lot of people out there who own something very similar.
This isn’t a bad thing, and I’m all for it as this allows multiple fans the opportunity to own their favourite tracks, and arguably the more that are minted, the more accessible it becomes for fans as the price will be driven down – simple supply and demand.
This approach to minting music NFTs is the one I can see becoming dominant, allowing users to build up their own collections at a resonable price point.
But as a fan, does it really matter to me whether I can get #347 / 1749 or #692 / 1749 – at this stage people will opt for whichever will set them back the least amount of ETH. The increased supply affects the perception of the value of each NFT.
This leaves artists to decide how many versions of their track to mint, making a choice between rarity and price.
This is where generative music comes in.
Generative music NFTs allow an artist to create multiple unique NFTs based off variations of an original track.
This allows them to retain a similar uniqueness to a 1/1 sale, but bringing the price point down to make them a more reasonable purchase.
In the same way as the PFP NFT projects like Cyrptopunks and BAYC have seen success, an artist can create far more value by making, to use that Brian Eno quote again, ‘music that is ever-different and changing, created by a system’.
Used effectively, generative music NFTs offer an interesting opportunity for those artists who want to try something new.
In order to get a clear picture about the potential for generative music NFTs, it’s also wise to be realistic about the obstacles that will be faced.
Utility – Profile picture NFT projects have seen great success because they can also fit quite nicely into peoples existing digital lives. Everyone is used to having a profile photo on their accounts, they can use this to actively display their NFT and signal to people that they are part of something.
Music doesn’t quite translate over in this way, and there are fewer places across the web where music NFTs provide the same utility. Until we see the digital world grow and offer more use cases, I expect there to be less demand.
Product / Market Fit – generative music NFTs are new, and there’s no way of knowing exactly what sort of audience is out there and how they will be received. In certain genres, such as electronic music, remixes are part of the scene, and very much encouraged.
It may be that different genres are more receptive to the idea of owning a unique remix of the track, and this won’t be as popular across the whole industry.
There’s an argument that with music, what’s important is actually the shared experience as opposed to rarity or uniqueness. This may be the case, but I think generative music NFTs will likely coexist as another medium through which an artist can express themselves.
As with most new concepts, it’s quite clear that we won’t know exactly how well received generative music NFTs are until the market develops. We shall have to wait and see!
Different Approaches to Generative Music NFTs
The approach to generative music will usually fall into one of two categories:
- AI / fully automated and randomized
- Human involvement with systemisation
On the one end of the spectrum you have music that’s fully automated, whether composed by AI, or a very specific set of rules. This can be generated at scale, and offers the advantage of being potentially infinite. A good example of this is the Audioglyphs collection.
On the other end of the spectrum you have generative music which heavily involves the artist in the creation process, and then uses a system to take this creativity and generate different variations of the initial artwork.
More, or Less Automation?
Fully automated music can be attractive to aficionados of experimental music who see the beauty behind this type of machine led approach, but there are some drawbacks.
Often it is much harder for fully automated rules to create music that is consistently audibly pleasing, and that has the same depth as that which has been crafted by an artist. The music is often more basic, and lacks the creative input that a real artist can bring to the table. Call it ‘soul’ perhaps, but whatever it is, full automation isn’t able to capture it yet.
Using a system to vary your audio also brings its own set of challenges. Good music is often a fine balance of complementary sounds, the more variations you bring into this process, the less likely that all the different combinations are going to work well together.
You also need to consider that the further you slide towards automation, the more you cut the artist out of the process. There’s a tradeoff of between the talent, and renown of an artist, and the rarity (difficulty of replication) of the audio.
When people invest in art, whatever the form, they invest into a story. You could argue that a machine, or an algorithm creating a song is a story in itself, but that story can only be told so many times before it becomes repetitive. It’s the personality behind the artwork that is important also.
It’s for this reason that I believe the most valuable type of generative music is going to be that where an artist interacts with a system to create unique versions of their artwork, all encapsulating the spirit of the original work, but with individual characteristics which make them unique.
How to Create Generative Music NFTs
For the average artist, their talents will lie in the realms of creating music, so creating generative music NFTs may seem beyond their reach, but things really aren’t too complex once you understand the process. You don’t need to learn to code, or understand AI, in fact to start with you can just do this very simply.
Below I will do my best to lay out a theorized version of how this would work.
Creating Multiple Stems
Most songs are made up of multiple layers of sounds / instruments (commonly known as stems), all designed to fit together in the right way to create something pleasing to the ear.
As long as these stems follow the same rules of the overall song (same key, same BPM), then there should be an element of interchangeability, for example you could use a different instrument / synth on the lead melody and the overall composition would still sound pretty similar.
By creating 4 different versions of the lead melody you actually create 4 different versions of the track.
When you apply this process across the stems, for example two different drum layers, two melodies, two pads, you can see how quickly you can end up being able to create many multiple variations of the track.
There does need to be some consideration when varying stems, as all parts need to be able to fit in with all possible variations of the track.
This is where the human aspect comes into play. Your ability as an artist to balance the difference stems to create something audibly pleasing each time is essential.
While I’ve done my best to get the concept over visually, it may be difficult to imagine what this might sound like, so I’ve pulled together a few audio examples to show you.
Using the above example I have put together a combination of the above stems with two different synths for the melody.
Here you can hear the two synths in isolation.
And here you can hear them combined with the other stems. Changing just one stem can create a similar but different version of the same track.
In this example the difference between stems is fairly noticeable, but you as the artist retain control over which elements you want to vary more and which you want to keep very similar.
There are no hard and fast rules here, you can vary things as much or as little as you want depending on the goals behind your project.
You might be releasing a collection of NFTs based on an existing song, in which case you might want to keep all the variations a lot closer to the original track, and only mint a handful.
Or you might be looking to create a larger project with no pre existing track, which means variations don’t need to sound as similar to each other.
Generating the variations
When you look at the generative art PFP projects, a lot of these are built from multiple different layers (e.g. eyes, hats, accessories) that are created in a photo editing software such as photoshop.
These layers are pulled together using a script which compiles them to create the final product. Rarity can be assigned to the different layers to make some characteristics more rare than others.
When it comes to generative music, there isn’t quite the same solution available yet (let me know if I’m wrong), but it’s a project I’m actively working on (developers reach out if you fancy getting involved).
For now the easiest, and probably most familiar place for an artist to create the different variations may be within their DAW (digital audio workstation) of choice, selecting the relevant combinations and exporting the master file.
This is going to be one of the considerations to factor into your process, for the time being the amount of effort involved in creating the variations may make you want to keep the collection size on the smaller side.
Minting Generative Music NFTs
There are a lot of opinions about the right way to host and mint NFTs – with lots of different approaches and platforms out there. You could use an easier but more centralized solution like Opensea or Rarible, or you could opt for a more decentralized approach using IPFS and your own custom built DAPP.
The reality is that your approach to minting is largely going to be determined by your level of experience with the technology, and what your end vision for the project looks like.
The process of minting your NFT’s is outside the scope of this document, you’ll find plenty of guides to minting online as well as plenty of advice on NFT launches. I would choose the process you are most comfortable with and adjust this as you become more familiar with the process.
Rather than take you through all the different ways of minting, instead I thought it would be useful to share a few considerations, which will determine how you approach your generative music project.
Generative Music Project Considerations
Pricing and Gas Fees
Pricing is a tricky area. A lot of people get into the web3 space because of a desire to make money as opposed to an actual passion for the technology and the community.
High pricing can be a barrier for new people coming into the space, and ideally you want to lower the bar for people who want to dip their toe in the water and invest in one of your NFTs. At the same time this new technology does offer artists a way to get rewarded for their art in a way that they haven’t been previously.
Your vision for your project will determine what value you want to set for your NFTs, and the market will ultimately decide if it’s worth it.
Gas fees should also be considered as part of the project. If you are at the high end in terms of pricing, people aren’t going to mind paying the gas fees to mint on the ethereum blockchain, but if you’re at the lower end it may put people off paying more in gas than the price of the token. If this is the case then it may be worth looking at minting on a sidechain such as polygon to keep the transaction fees down.
Blind / Masked Minting
Imagine being able to own a song by your favourite artist that hasn’t been heard by anyone else in the world. This could obviously be quite an attractive proposal to an enthusiastic fan, and it’s generative music can help make this happen.
When launching an NFT drop, one of the things to consider is whether you do a blind mint. With a blind / masked mint, the purchaser doesn’t know what variation of the track they will be getting before they mint it.
A good example of a similar comparison is loot boxes in video games, where the user doesn’t know what’s inside until they open it.
It’s this process in the generative art world that allows 10,000 NFTs to be minted for the same price, but the resale value of each NFT varies based on the rarity of some of the attributes.
A blind mint would mean that the stem variations are combined only once the NFT has been minted, so the purchaser wouldn’t actually know what the track they were going to get sounds like (although it shouldn’t hopefully stray too far from the original if you have one).
This process has pros and cons, while some people may like the anticipation of not knowing what their actual track will sound like, others may appreciate the ability to browse the different variations before purchasing and select one that’s right for them.
The major draw that blind minting has is that it creates a variation that no one else will have heard.
The majority of music NFTs consist of audio paired with visuals. Sometimes this is static cover art, sometimes it’s a video. I would strongly recommend considering visuals to go with the audio as it does complete the experience and helps the NFT stand out more on marketplaces.
Of course this all depends on your specific project and what the tech landscape looks like. At the moment the majority of NFTs are visual, and it follows that the way people display these NFTs is also visual.
Going forward this may change as audio NFTs become more popular. It’s likely that you will be able to link up your collection and play it in playlist form, and as the industry develops it may reach a stage where audio only NFTs are more desirable, but for now my advice would be to pair the audio with visuals of some sort.
Music files are data heavy, and storing audio on chain isn’t an option, so you will need to find another way of storing your audio. On a side note there are a few interesting projects who are making developments in this area, such as Arpeggi Labs, so be sure to keep an eye on them.
If you are using a platform such as Opensea to host your project this isn’t as much of a concern as they will have an allocated storage amount per file, but if you are uploading them to IPFS yourself keep in mind that storing lots of large files will soon add up in terms of storage, and you may find yourself needing to pay out for this.
Number of Variations
By varying the stems you can easily see how you can end up with vast amounts of different variations. The more there are, the more they are going to vary from the original song.
It also means more work to create each of these files.
Your number of variations can also impact the subjective value of the NFT, and you won’t know until you start making them.
My thoughts would be to start small and manageable. Focus on quality and get a feel for the process. If you need to you can then ramp it up from there.
Copyright / Ownership
Make sure you have ownership of the music locked down. If you’re signed in any way you’ll need to make sure you know what you can and can’t be doing with the music – the last thing you want to do is end up with headaches or legal action down the line.
For most unsigned artists who are in complete control of their music this shouldn’t be an issue.
User Adoption & Promotion
If you have a big audience on existing streaming platforms and are used to a lot of engagement from your fans, you may not necessarily see this carry over to your NFTs.
As with all new technology there’s a learning curve, and only a small percentage of your audience are likely to be active in the space. Not only will you need to get past this learning curve yourself to really understand what’s going on – you’ll need to consider a certain level of promotion to get people aware of your project – but the good news is you’ll still be early to the space.
After touching on promotion, it’s also worth noting collaboration. Collaboration is an extremely common marketing approach in the music industry, and we can see this spill over to music NFT projects also. A lot of projects are either collaborations between artists, or perhaps between an artist and a programmer.
Multiple people working on a project can create a real synergy, but one of the things to be aware of is how you split the income you make from the NFTs. Not all marketplaces currently cater for this, so it would either involve some level of trust, or the development of a smart contract that can split this for you.
Another bit of advice is to not get hung up on selling your collection straight away. Even some of the biggest artists in the world have launched music NFT collections that haven’t sold out.
Most artists will go through a period of little or no recognition at the start of their careers, and there’s no reason why this should be any different approaching web3.
If you go into this process with expectations of making a fortune and blowing up overnight you are setting yourself up for disappointment.
If you instead view the whole thing as a learning experience – a way to help contribute to the culture, and a way to push your boundaries, then the whole experience will be much more meaningful.
Where To Go Next…
Well done on making it this far through. Hopefully by now you should have a better understanding of generative music, how this can be applied to NFTs and how to start experimenting with the process.
So what should you do next?
You’ve got a couple of options. Either you can dive straight in and start playing around creating your own tracks and minting your own NFTs.
Or follow along with the music NFT scene and see the sort of impact they make.
If you’re an artist or developer who’s interested in this area then feel free to get in touch.
I’ll be aiming to go through the process and I’ll share what I’m doing and so you can get an idea about the level of interest out there. Sign up to the newsletter and I’ll keep you updated.
Either way, I hope you’ve found this deep dive interesting and look forward to seeing what comes out of the generative music NFT space in the future.
Also if you liked this article, don’t forget to share it on you socials 🙂