AI Music Will Break The Internet

Over 200 artists, including some of the biggest names in the music industry like Billie Eilish, Nicki Minaj, Zay Malik, and Katy Perry, recently made a joint appeal to technology companies. They urged these firms to stop undervaluing music and using AI in ways that infringe on artists' rights.

We call on all AI developers, technology companies, platforms and digital music services to commit not to develop or deploy AI music generation content or tools that undermine or replace the human artistry of composers and artists or deny us fair compensation for our work

This raises important questions about the current state of AI music generation and whether it poses a genuine threat. Let's explore this topic further.

The current state of AI in music

AI tools for generating music and sound effects have existed for some time now. Major tech companies and startups alike have released their own AI models in this domain.

• Google introduces Lyria, an AI music generator.
• Microsoft introduces Muzic, a research project on AI music generation.
• Stability AI released Stable Audio 2.0
• Udio and Suno drop their AI tools that generate music with vocals.

These are just a few examples of the AI music generation tools currently available. They are able to produce high-quality music based on simple text descriptions provided by users.

However, the pace of AI advancement in the audio space has accelerated dramatically in recent times. AI has now become sophisticated enough to articulate words and sing complete original tracks.

To clarify, AI is now capable of writing lyrics, composing music, and performing the vocals in astonishingly realistic and emotionally expressive voices. This is a truly remarkable development.

Udio AI

Udio functions similarly to an image generation tool. Users provide a description of the song they want, and the AI generates an original track within seconds.

The process of creating a song in Udio is quite straightforward. After creating an account and logging in, users simply describe the desired song. They have the option to manually input lyrics or let the AI generate them.

Once the user hits the create button, the AI gets to work. For each song request, the AI generates two versions, likely to provide users with options for the direction of further edits.

The generation process is quite fast, taking less than a minute to create a 30-second song about "artificial intelligence" in this example. The results can be accessed under the "My Creations" page.

Udio even generates album art and titles for the songs it creates. It's truly mind-blowing. Here are the full lyrics for the sample song:

We can code a world inside the screen
Feel the algorithm’s rush
Watch a digital dream come to life
Chrome and glow
I know we’re getting smarter
Don’t slow it down
With every click and every code we’re soaring
Breathing life into the circuits we’re exploring
A.I
whispers wisdom, like a breeze
Changing the horizon, into infinite seas
[Pre-Chorus]
We’re in sync with synthetic minds
Advancing, entwining like digital vines

Listen to the final result here:

If 30 seconds isn't sufficient, users have the option to extend the song by an additional 30 seconds. They can specify whether to place the extension at the beginning or end of the track.

How much does it cost?

Udio is currently in beta and generously offering its services for free, with a limit of 1,200 songs per month. They haven't yet announced what the pricing model will be or when it will be implemented.

So if you're curious to explore its capabilities, now is a good time to try it out while it's still free.

Suno AI

Suno AI is a new AI-powered music generator that is gaining significant attention in the music industry. Similar to Udio AI, users can generate instrumental tracks, vocals, and even lyrics, all from a simple text input.

Suno AI’s music generator offers three plans:

1. Basic (free): 50 daily credits, non-commercial use, shared queue, 2 concurrent jobs.
2. Pro ($10/month): 2,500 monthly credits, commercial use, priority queue, 10 concurrent jobs.
3. Premier ($30/month): 10,000 monthly credits, commercial use, priority queue, 10 concurrent jobs.

Both paid plans offer a 20% discount with annual billing and the option to purchase additional credits.

How’s the audio quality?

I am not an expert in music, but if I am going to make a judgment based on what I hear, I would say that it’s really, really good.

Let’s talk about copyright issues

The current state of copyright law and regulations surrounding AI-generated music is still evolving, with ongoing debates about how to protect the rights of artists and composers in this new landscape. It's somewhat of a Wild West situation, with no clear rules or guidelines in place.

Udio does seem to have some awareness of copyright issues. For example, it will not allow users to request that it sing Frank Sinatra's "My Way", displaying an error message stating that it does not generate artist likenesses without permission.

However, there doesn't appear to be anything preventing users from taking the lyrics to "My Way" and having the AI compose a drum-and-bass track around them if they so choose. It's a bit of a grey area, and it remains to be seen how copyright law will adapt to these emerging technologies.

Should we be scared?

As mentioned in the introduction, many artists and composers have voiced concerns about the potential impact of AI on their livelihoods. They contend that AI-generated music could devalue their work and deprive them of fair compensation for their creativity. This is a legitimate concern that warrants serious consideration.

However, others argue that AI-generated music could open up new avenues for creativity and innovation. For instance, AI could assist musicians who are facing creative blocks or other challenges.

Black Eyed Peace singer will.i.am quoted an example audio clip from Udio on X, saying this is the best tech on earth.

Currently, there are no laws specifically addressing AI-generated music. However, some legal experts suggest that existing copyright laws could potentially be applied to AI-generated music, depending on the specific circumstances.

For example, if an AI model is trained on copyrighted works, the resulting music could potentially be considered a derivative work, which would require permission from the original copyright holder. It's a legal minefield, and it's unclear how the courts will interpret these issues.

What’s your take on an AI music generator?

Does it impress or scare you? I’m a bit of both, to be honest.

Final Thoughts

The ability of AI music generators to add vocals is astounding, but understandably, not everyone is thrilled about it.

The use of AI to generate music is a complex issue that raises important questions about copyright, creativity, and fair compensation for artists and composers. There are no simple solutions, and we can expect to see many legal battles and debates in the coming months.

However, one thing is certain: AI is unstoppable. Currently, one of the main drawbacks of AI-generated music is its lack of emotional depth. But imagine a future, just a few years from now, where AI becomes so advanced that it becomes impossible to distinguish between a human singer and an AI-generated voice.

Apps could be developed that allow users to request songs on-demand based on their current mood. Or perhaps users could ask the AI to have their original compositions sung in the style of Taylor Swift or Beyoncé. Is this a future you'd like to see? I'm very interested to hear your thoughts on this.