Meta’s new Generative AI tool that tan create music from text


The rise of generative AI tools has already enabled the creation of text and visual content, extensively reshared across the web. However, the realm of audio and music generated by AI systems has also emerged as a potential disruptor, albeit facing resistance from the music industry.

Notably, there have been instances of fully AI-created music gaining popularity, such as a viral track featuring Drake and The Weeknd, which involved no contribution from the artists themselves. This points towards future disruptions within the music industry, although the industry has expressed strong opposition to such developments.

Nevertheless, there are various ways in which generative AI tools can facilitate novel forms of music creation, and this is precisely the focus of Meta’s latest generative AI model called ‘MusicGen.’

MusicGen utilizes text or melody prompts to generate entirely new music, drawing inspiration from samples of songs and instrument styles embedded within its generative elements. By specifying the desired type of track or even humming a tune, MusicGen produces variations of audio as outputs.

The MusicGen model has been trained on an extensive dataset comprising 20,000 hours of music, including complete tracks and individual instrument samples, offering a wide range of inputs for AI-generated compositions. While it is not yet widely available, a demo can be accessed to get a glimpse of its capabilities. The potential implications for the future are significant, as it provides a new avenue for creating original music, potentially transforming the approach of musicians, marketers, and others.

However, like all AI creations, MusicGen may encounter legal challenges, especially if the music industry intervenes. The music industry has already formed dedicated teams to identify and address copyright violations online, encompassing not only simulations of renowned artists but also unauthorized sampling of their owned content and tracks. This could potentially lead to the implementation of regulations to prevent systems like MusicGen from operating, although the usage of distinct samples could make legal action challenging.

Inevitably, AI-generated songs could become mainstream hits, even without listeners realizing their origins. Nevertheless, the potential uses of tools like MusicGen extend beyond mere replication, offering new possibilities for musical expression in various forms.

Whether this is the future we desire or not is inconsequential, as it is undoubtedly on its way. Eventually, it will pave the way for diverse individuals to create their own music for a range of purposes, opening up new avenues of creative exploration.