Google Just Developed An AI Model That Creates Music From Text Prompts — But The Company Won't Be Releasing It Anytime Soon
- Google has announced that they have developed an artificial intelligence robot that creates music based on text descriptions.
- This technology will not be released to the general public due to technical issues and risks.
In the race for AI, Google has announced that it has developed a bot that makes music from text phrases, but isn't expected to use it any time soon.
In a study published Thursday, Google researchers described MusicLM as "a model that produces high-quality music from text descriptions, such as relaxing violin melodies with distorted guitar riffs."
“We demonstrated that MusikLM can be conditioned by both lyrics and melodies, as it can melodically change whistling and crashing melodies according to the style described in the footnotes,” the document states.
Based on the analysis, users can enter descriptions such as “catchy jazz track with saxophone leads and catchy solos” or “90s Berlin techno with deep bass and powerful kicks” and get results accordingly. A similar example posted on a Google Github page displays audio relevant to such a question.
MusicLM's debut comes amid the meteoric rise of chatbot OpenAI ChatGPT, which led to Google issuing a "code red" that the New York Times reported in December was "akin to sounding a fire alarm" for the tech giant.
To stay competitive, the company is accelerating the rollout of 20 new products, including a version of Google search with AI chatbot capabilities, according to the Times.
However, Google has stated that they have no plans to release MusicLM to the general public, citing several risks, including bugs that could lead to informational and cultural misrepresentation, technological interference, and indeed "the potential for distraction from creative content".
Based on the analysis, identifiable songs were found in approximately 1% of the sample, indicating potential copyright infringement.
“We strongly stress the need for further work to address the risks associated with producing this music; We currently have no plans to launch that model," the studio said.
The study also pointed out the limitations of the technology, including the use of reflections and the time spent in text queries, and sound quality. In the future, the researchers plan to work on "high-level modeling of song structures such as intros, verses, and chorus."
