Is Google’s music AI generating the sounds of the future?
The world of music production is constantly evolving, and now, with the help of Google’s AI Music Generator, MusicLM, it’s taking a significant leap forward. This new technology can create music from scratch, allowing producers and songwriters to experiment with sounds and styles that were previously difficult to achieve.
How Google’s MusicLM works
Google’s AI music generator is powered by a deep learning algorithm that can generate music using text-based inputs. According to The Verge, it can create music in any style or genre and adjust the tempo, key, and other elements to create a unique sound.
The algorithm is based on an artificial neural network, a machine learning system that takes in data and uses it to make predictions. In this case, the algorithm uses various musical elements, such as notes, chords, and rhythms, to generate a unique piece of music. But since Google’s programmers used existing, copyright-protected music, it’s estimated that about 1% of the output is direct copies of existing songs.
Cue the outcry from musicians! And as The American Genius reports, rightly so.
The problem with AI and intellectual property laws
AI technology and intellectual property laws have come into conflict in recent years, with issues such as who owns machine-generated works. If a machine creates a painting or a song, who owns the copyright? AI algorithms can read and analyze large amounts of data, potentially infringing on individuals’ trademarks and copyrights.
In the realm of AI image creation, there have been several notable examples of copyrighted photographs obviously serving as a source. Getty Images recently filed a lawsuit against Stability AI for this exact type of misuse.
Because of these copyright
infringements questions, Google is not yet releasing the music-generating technology to the public.
There are practical uses, however, that do not necessarily pose copyright issues. For example, if the algorithm is fed a prompt for a piece of classical music, it will be able to create a piece with the same structure and feel as other public-domain classical pieces. This generated music could be used in an independent film or online streaming content.
AI music could help creators produce music more quickly and have access to a wider range of sounds on a low budget, but until copyright concerns can be completely eliminated, the financial risks outweigh any benefits.
Challenges and downfalls of AI music Generators
While Google’s MusicLM could prove to be an incredible tool for producers, it does have some limitations. While it is capable of creating music that is similar to a particular artist or genre, it is not able to create something truly original. Melodies, harmonies, and even vocals are all based on someone else’s previous recordings.
Another challenge is that the algorithm needs to be able to capture the subtle nuances that make human-made music so compelling. Right now, that’s a big downfall with the output. Much like AI-generated text, a human edit or remix, in this case, is essential to make AI music come to life.
And, of course, mirroring worries from writers and artists about other generative AI like ChatGPT and Midjourney, there are concerns by flesh and blood musicians about whether or not this tech will ultimately take away paying jobs for musicians and composers.
Try an AI music generator now
While you can’t fully get your hands on Google’s MusicLM just yet, you check out other promising AI music generators today.
SingSong is a tool that produces instrumental music that follows the pitch and rhythm of vocals provided as input. This can be of great use to professional and non-professional musicians who want to create music featuring their singing.
The developers of SingSong employed the most recent advances in musical source separation and audio generation to bring this project to life, and the results are pretty good. They took a vast collection of music audio and ran a leading source separation algorithm to generate sets of vocals and instrumental sources that are in sync.
Are we ready for AI music?
It’s a fair question. But with other generative AI options such as ChatGPT and Midjourney more popular by the day, it’s hard to imagine a world where audio-generating AI isn’t all around us.