My partner Filip and I have been experimenting with AI music through our sister venture Tuney since 2020. As the topic of "generative art" continues to grab headlines, I wanted to share everything we've learned so far and how it will impact the advertising production industry going forward.
How far along is the technology behind "AI Music"?
Technically, computers have been "composing" music since the 1950s, but in recent years a breed of companies have been developing algorithms that generate musical audio out of thin air the same way DALL-E2 and Midjourney generate visual art.
Early experiments produced "decent" results considering it's computer-made music with no human involvement, but nothing that could be remotely passable for use in branded content. Early companies to experiment with generative music have been Jukedeck (acquired by Bytedance, TikTok's parent company), Amper (acquired by Shutterstock) and AIVA.
In the last few years, the space has heated up significantly. The quality of the music output is getting better, and new methods are being used to combine human components with AI-generated ones as a means of improving overall quality.
Will AI music replace "library music"?
Music created with 0% human involvement is still a ways off from being usable in commercial content beyond the odd novelty use. But music with as low as 10% human involvement is producing some incredible results.
Even our very own AI music project Tuney is capable of producing high quality music, including vocals, because it uses human-made musical building blocks as a starting point (you can listen to some examples here).
There's a ton of library music out there so even if AI does a great job imitating it, most creatives still prefer a human-made song. At the end of the day, library music is pretty affordable. If you go the AI music route and you're not trying to save money, I assume you'd go for the PR factor. For that, you want the AI component front and center, otherwise no one will know you're using AI music in the background.
Will AI music replace "custom music"?
This is where AI could provide some serious value. Customizing and personalizing music still requires a bit of human effort and is a lot more expensive. While we don't personally like the idea of AI replacing composers (after all, we're music producers ourselves), there are some great use cases for AI when human production doesn't make sense from a timing or budget perspective.
For example, Tuney can generate a track version of any length and place the peak of the music anywhere along the timeline of an edit. If you need cheap cutdowns and adaptations without a budget for a composer or music editor, the platform can make them for you.
Pretty soon, custom vocals will also be available and all you'll need to do is type in the words (the AI will even take prose and make it rhyme for you before turning it into perfectly-sung vocals).
Will AI write the next Billboard 100 hit?
People keep asking this question. It could be possible, and sooner than you think. Some companies in Asia have already done it, albeit with mixed results.
I personally believe that even if the technology is capable of delivering the perfect song instantly, most music fans would prefer to pay attention to the music that artists create because it speaks to our emotions and gives us a human story "behind the music".
What about virtual artists?
Virtual artists have been grabbing headlines in recent years, from Lil Michela, to FN Meka, to Authentic Artists. In most cases, a heavy dose of human intervention is involved in the final product. In most cases, the voices are human and the production is manual.
Gorillaz are arguably the most successful "virtual artist" out there, and they've been at it long before any significant advances in AI music were made.
At the end of the day, virtual artists are going to be more about character development and storytelling than cutting edge technology used to produce the music they make.
How can brands and agencies use AI music creatively in 2023?
This is the fun part! So far we've been discussing music production being automated by AI, but what about using AI music technology as the cornerstone of an interactive or experiential campaign?
This is where the current AI music technologies truly shine, and where brands can enable unforgettable music experiences for consumers and fans.
Some projects we're already exploring with brands:
A remix app that lets fans mash up their favorite songs
A music maker app that lets creators make music using AI
A text-to-song experience, where users can type lyrics to generate a finished song
An app that creates a personalized music feed for meditation, fitness or focus
An entire generative music library, where each track features the brand's sonic melody