AI's Sonic Disruption

When starting to think of music for advertising films, one of the first questions that comes to mind usually is: “Sync, library or bespoke?” Now it’s looking increasingly likely that the question will become: “Human-created or AI-generated?”

The music industry is experiencing what some may call a revolution, others an upheaval. It’s all powered by AI. While creative fields like design and writing have already seen dramatic shifts thanks to tools like Midjourney, Runway and ChatGPT, music is now stepping into the spotlight. With AI-driven technology, the way we create, produce and even think about music is evolving at lightning speed, whether we like it or not. As a music supervision company, we are at once concerned about the disruptions it brings to our partners while seeing many operational and creative opportunities at the same time. 

In just a few years, AI has made its mark on the advertising music industry in ways we could barely imagine. From voice cloning to similarity search and stem separation, these innovations are reshaping the industry. But the real elephant in the room? Generative AI. Tools like Suno AI and Udio can now create entire songs, with vocals, all from a simple text prompt. It might be a new era where anyone can generate music with the push of a button.

Yet, with all this innovation comes a wave of challenges. Generative AI in music is not just a cool new toy - it’s at the centre of fierce legal battles. As AI-generated tracks start to make waves, the industry is bracing for a showdown over copyright, ownership and the future of music.

So, what does this mean for the world of music? Let’s dive into the potential pitfalls, the legal implications and what the future of music for advertising can look like.

What’s the Deal with AI-Generated Music?

The generative AI debate falls into two camps. Some say that it democratises music creation, allowing anyone to potentially become a music creator in their own right. But the naysayers decry generative AI as a mere plagiarism tool, as it is trained on millions of pre-existing songs, including existing melodies and lyrics, before replicating them.

This comes as Udio and Suno admit that they have trained their AI off unlicensed music. The former might claim that they identify and block tracks that too closely resemble popular tracks, but this only protects already-famous artists and poses damage to lesser-known artists. It also begs the question: 

Is This Even Legal?

While image and text content is owned by millions of different entities, the entire history of music is essentially controlled by a handful of organisations, the major labels and publishers, as well as independents represented by their industry associations, giving music rights owners a significantly stronger position than its other industry peers when it comes to the question of training data.

AI companies may claim fair usage or “transformative” use, but the intellectual property holders are pushing courts to agree that the underlying content itself is trained on swathes of copyrighted data. This comes in the form of unlicensed music and artist likenesses, the usage of which requires permission and remuneration.

In the EU, the recently passed AI Act states that the training of AI models on copyrighted material requires permission from rights holders. In the USA, multiple challenges to AI material already exist, but this topic has yet to reach the Supreme Court, opening up the current legal mess of uncertainty and confusion. It is still unknown how AI will be legislated in the future, especially when considering the deep pockets bankrolling the current boom.

Music and AI: More Than Just Generative

In addition to generative AI, many other uses of AI can put music licensing and production ahead of the pack. For example, AI can now cleanly separate audio from instrumentals, something extremely important to properly mix older songs without stems into a spot.

Voice cloning gained attention in a 2023 Volkswagen ad in Brazil, where the late singer Elis Regina was digitally revived to perform a duet with her daughter, Maria Rita—something previously impossible as Elis passed away when Maria was four years old. This was a seismic event, with Brazil’s First Lady Rosângela Lula da Silva among millions of Brazilians moved to tears by the artificial restoration of a national icon.

Another growing functionality is similarity search, whereby one can search for existing tracks that have similar sonic qualities to others. This can be useful when looking for something that evokes a certain vibe and characteristics, especially when looking for more affordable licensing options.

Free text searches are also on the rise and can be a handy tool for music supervisors or anyone else looking for music. The free text prompts can be specific about genre, pace and so on, but can also be more broad, such as ‘’songs that could be in a Tarantino movie.”  Through its understanding of music in context, the AI is able to provide search results that are not tagged or sorted by any human-generated metadata.

These examples show the creative and interesting sides of using analytical AI as well as some forms of generative AI in music which can solve a lot of problems and save time and costs. 

It is clear that AI is here to stay. These tools range from supportive to extremely disruptive. The question is not if, but how the industry will adapt to the existence of these new tools when it comes to providing music for advertising films. 

Human or AI? 

From our perspective as providers of music for advertising, now that generative AI can create infinite amounts of music at a fast speed and low cost, competition will increase, affecting everyone. Here are some key insights we gathered from hundreds of conversations with clients, agencies, labels, publishers, libraries, music houses and AI researchers: 

• Legality: Currently Gen AI music is not legally safe to be used in any professional production and clients are well advised to prohibit composers from using Gen AI samples until further notice. However, in the medium term, AI music that is legal and fit for use is on the way; it is only a matter of how and by who it will be provided. Current providers are still like Napster, but a Spotify equivalent will be coming.
• The Ownership question: As AI-generated songs will inevitably gain recognition and in some countries are already entering the charts, ownership battles will emerge over who truly owns the music - the prompter, the original creators of the training data, platform providers or a combination of them all. Early proposals for solving training data remunerations are already being floated in the market, but it will take some time until we see generally accepted distribution models and robust tracking technologies. 
• Social Media Culture: Platforms such as TikTok have a culture of sped-up, slowed-down, remixed and heavily modified versions of music already. Quality is not king, instead, user interaction is king. This will likely serve as a channel for normalising as well as promoting AI music. 
• Competition: AI's ability to create cost-effective music puts pressure on artists to highlight their unique value, whether through quality, precision or fame, as the competition intensifies.
  
• The Role of the Composer: Music makers may face a diverging career path of becoming either well-trained prompt engineers or distinguishing themselves as one-of-a-kind artists in their handcrafted niches.
  
• Scarcity and Value: As AI floods the market with music, the value of hits and iconic artists will most likely increase, just as many luxury goods thrive on exclusivity and restricted access.
  

Conclusion

Lawsuits are just the beginning for generative AI in music, with many headaches to come as the genie is already firmly out of the bottle. For one thing, a tidal shift is on the horizon, and while it's unclear exactly what will happen, we hope this article has provided you with insights into where we think it will go. 

But no matter which technology is being used: an electric guitar, a sample or generative AI, the biggest value often comes from a simple creative idea. And the role of music supervisors and music houses will still be to come up with and identify the right ideas at the right time.

Speaking of the right idea at the right time, when it comes to music, please feel free to get in contact with us at Tracks & Fields!