'This is bigger than ChatGPT': Google creates 'MusicLM,' text-to-music AI

For now, the tech giant has decided against making the AI trained on a dataset of 280,000 hours of music publicly available due to ethical concerns.
Baba Tamim
Representational image: Text-to-music AI.
Representational image: Text-to-music AI.

Vladyslav Otsiatsia/iStock 

Google has created a new artificial intelligence (AI) system called MusicLM that can produce music in any genre from a text description.

However, the tech giant has chosen against to make the system publicly available due to ethical issues, according to some media reports.  

"Whoa, this is bigger than ChatGPT to me. Google almost solved music generation, I'd say," Keunwoo Choi, an AI Scientist at Gaudio Lab, Tweeted on Friday.

Although MusicLM wouldn't be the first generative AI system for music, it is the first to create songs with "high-fidelity" and complicated composition.

The algorithm can produce songs that make sense for descriptions of "substantial complexity" after being trained on a dataset of 280,000 hours of music.

The system can build on existing melodies, whether they are whistled, hummed, sung, or played on an instrument. 

It can also take a series of sequentially written descriptions and turn them into a musical "story" or narrative, according to Google researchers. 

Additionally, MusicLM can be directed by a combination of a picture and a caption, or it can produce music that is "played" by a certain kind of instrument in a particular style.

Although the system can technically synthesize vocals, the results are far from ideal and have problems like distorted samples. 

The copyright issues

The key issue for Google is the potential for MusicLM to use training data that contains copyrighted material in the songs that are produced.

In an experiment, the researchers discovered that one percent of the music the system produced directly copied the songs on which it was trained. 

This figure is high enough to make the business hesitant to release MusicLM in its current form, noted a TechCrunch report on Friday. 

The researchers highlighted the necessity for more future effort in addressing these hazards related with music generating and underlined the risk of potential creative content misappropriation linked with the use case.

However, some people are still awwed by AI-audio bites released by the Google.

"Impressed to see the quality of autogenerated vocals has gone way up! Sounds real but in a foreign language," wrote a Twitter user. 

Not the first time 

It's not the first time that AI-generated music has given rise to legal concerns.

Jay-record Z's company filed copyright complaints against Vocal Synthesis in 2020 after the YouTube channel used AI to produce Jay-Z renditions of songs like Billy Joel's "We Didn't Start the Fire."

The films were initially taken down, but YouTube later decided the takedown requests were "incomplete," so they were put back online.

In a whitepaper, Eric Sunray, a current legal intern at the Music Publishers Association, claims that AI music generators like MusicLM are infringing on the copyright by weaving "tapestries of coherent audio" out of the compositions they consume during training, the TechCrunch report cited.

It's pertinent to ensure that AI-generated music may be used in a way that is fair to both composers and users, the industry must address these ethical and legal challenges as AI technology develops.

However, it may take some time before there is some clarity on how courts will rule about the usage of AI-generated music.

You can find the Google study titled "MusicLM: Generating Music From Text" here and A-generated text-to-music samples here

Add Interesting Engineering to your Google News feed.
Add Interesting Engineering to your Google News feed.
message circleSHOW COMMENT (1)chevron
Job Board