Google introduces AI system that generates music from text descriptions

Representative Image [Photo: IANS]


Google has introduced a new AI system called – “MusicLM” that can generate high-fidelity music in any genre given a text description.

However, the company is fearing the risks, and has no immediate plans to release it, reports TechCrunch.

Google isn’t the first to try generative AI for songs, as there have been other attempts, including Riffusion, an AI that composes music by visualising it, as well as Dance Diffusion, Google’s own AudioML and OpenAI’s Jukebox.

But due to technical constraints and limited training data, none of them has been able to produce songs that are particularly complex in composition or high fidelity.

MusicLM is perhaps the first that can, said the report.

MusicLM was trained on a dataset of 2,80,000 hours of music to learn to generate coherent songs for descriptions of “significant complexity” (eg., “enchanting jazz song with a memorable saxophone solo and a solo singer” or “Berlin ’90s techno with a low bass and strong kick”), as described in an academic paper by US-based Cornell University.

Meanwhile, Google announced that users can now lock their incognito session when they leave Chrome on Android.

“You can require biometric authentication when you resume an Incognito session that was interrupted,” the tech giant said in a blogpost.

Earlier, this feature was available on iOS devices, but now it is rolling out to Android users too.