Last March, a “ChatGPT for music” called Suno was released. Suno uses generative AI to generate realistic songs on demand from short text prompts. A few weeks later, a similar competitor, Udio, appeared on the scene.
I've been working with a variety of creative computing tools as a researcher and producer for the past 15 years, and I've been shocked by the recent pace of change. As I have argued elsewhere, the view that AI systems will never make “real” music like humans should be understood more as a claim about social context than about technical ability.
“Of course, you can create expressive, complex, naturally structured, virtuosic, and original music that can stimulate human emotions, but AI cannot. appropriate “Music” can easily start to sound like something out of a Monty Python sketch.
After playing Suno and Udio together, I got to thinking about what exactly it is they're changing, and what it could mean for not only the way professional and amateur artists create music, but the way we all consume it.
Expressing emotions without feeling them
Generating audio from a text prompt itself is not new. However, Suno and Udio have made notable progress. It generates song lyrics from a simple text prompt (using a text generator similar to ChatGPT), feeds them into a generative speech model, and integrates the generated music with “vocals” to produce them. Consistent song segments.
This integration is a small but remarkable achievement. The system is very good at creating coherent songs that expressively “sing” (I personify).
The effects can be surprising. I know it's AI, but voices can still have an emotional impact. When the music plays a perfectly executed end-of-bar pirouette into a new section, my brain gets that little spark of pattern-processing joy you get when listening to a great band.
To me, this highlights something that is sometimes missed about musical expression. AI doesn't have to. experience Dealing with emotions and life events successfully express With music that resonates with people.
music as everyday language
Like other generative AI products, Suno and Udio are trained on a vast amount of existing work done by real humans, and there's a lot of debate about the intellectual property rights of those humans.
Nonetheless, these tools could mark the beginning of a mainstream AI music culture. They offer a new form of musical engagement that people want to use for their own enjoyment, explore, play, and actually listen to.
AI capable of “end-to-end” music creation is arguably a technology for music consumers, not music creators. At the moment, it's unclear whether Udio and Suno's users are creators or consumers, or whether that distinction is useful.
A long-observed phenomenon in creative technologies is that as they become easier and cheaper to produce, they become used for more casual expressions. As a result, the medium evolved from an exclusive high art form to an everyday language. Think about what impact smartphones have had on photography.
So imagine, with minimal cost and effort, you can send your father a professionally crafted song about him on his birthday in the style he prefers – a modern birthday card. Researchers have long considered this eventuality, and now we can do so. Happy birthday, Dad!
Is it possible to create without control?
Whatever these systems have achieved, and what they may achieve in the near future, they face a notable limitation: lack of control.
Especially in music, text prompts are often not as good as precise instructions. So these tools are suitable for a kind of blind exploration through the space of possibilities, but not for precise control. (That's not to diminish their value. Blind search can be a powerful creative force.)
As a working music producer, looking at these tools makes things look very different. Udio's info page says “Anyone with a song, lyrics, or interesting idea can now express themselves through music,” but I don't feel like I have enough control to express it. myself With this tool.
As with samples and field records, you may find them useful for seeding raw materials for manipulation. But when I try to express myselfI need control.
With Suno, I had fun finding the creepiest, darkest techno grooves I could get out of it. The result is something you'll definitely want to use on the track.
But I also knew I was willing to listen. I didn't feel pressured to add anything or manipulate the results to add my mark.
And many jurisdictions have declared that copyrights will not be granted just because something was created through AI.
First of all, the outcome is highly dependent on everything that went into AI, including the creative work of millions of other artists. Surely, you No work of creation was done. You just asked for it.
Between production and consumption, a new musical experience in a no-man’s land
So Udio's declaration that anyone can express themselves through music is an interesting provocation. People who use tools like Suno and Udio can be considered more consumers of music AI. experiences Than music AI creator factoryOr, as with many technological influences, they may need to come up with new concepts for what they do.
The shift toward generative music may be of interest to current forms of musical culture, just as the era of recorded music has seen the decline (but not death) of orchestral music, which was once the only way to hear complex, tonal, and loud music. . The explosion of engagement and interaction with these new types of music culture may lead to a decline in engagement with traditional music consumption of artists, bands, radio, and playlists.
It's too early to tell what the impact will be, but we need to pay attention. Efforts to defend the protection of existing creators' intellectual property rights, an important moral rights issue, are part of this equation.
But even if it succeeds, I believe it will not fundamentally solve a potentially explosive cultural change, and claims that such music may be inferior have long since been as effective in preventing cultural change as they have historically been, as have techno or even jazz. There wasn't. Government AI policies may need to go beyond these issues to understand how music works socially and ensure that our musical culture is vibrant, sustainable, enriching and meaningful for both individuals and communities.
This article is republished from: conversation Under Creative Commons License. read original article.
Image credit: Pawel Czerwinski / Unsplash