USA: The impressive MusicLM AI model developed by Google can generate various genre-based audio samples by simply following a text description.
MusicLM can take multiple text descriptions and follow them in sequential order to create a sweet story or narrative-style music. It can also build on tunes you already know and hum, whistle, sing or play along. Let's learn about its potential in more detail.
A Generative AI system for music is called MusicLM from Google. The tech giant claims that the model was trained on a dataset of 280,000 hours of music to learn how to create coherent songs from descriptions of "significant complexity".
Also Read: Centre Notifies Grievance Panels to Look Into Social Media complaints
It can maintain a constant music frequency of 23kHz during a clip, which can last anywhere from a few seconds to several minutes.
Google's MusicLM can create both short and long pieces of music in almost any genre. Even audio in the form of hums, whistles, melodies and choruses can be understood as music.
Also Read: Twitter rules to change from Feb 1, users will get this facility
All major music genres are supported by the AI model, including Jazz, Techno, British Indie Rock, Hip-Hop, Reggae, Folk and more. But neither the company nor any Indian musical subgenres such as Hindustani classical or Carnatic are mentioned.
To be perfectly honest, MusicLM isn't perfect yet. The quality of some of the generated samples is distorted. Specifically, the AI model can mimic vocals, but the sounds it produces are made up of nonsense words, which have been added to the musicians' synthesized voices to create them.
Also Read: Google modifies the Indian Android ecosystem significantly
Google researchers found that 1% of the music produced contained exact copies of the songs the system was trained on. Copyright would be infringing.