Nvidia Unveils Fugatto: Redefining Music And Audio Creation With AI

Nvidia has unveiled Fugatto, an advanced AI model designed to generate and transform music and audio, aimed at creators in the music, film, and video game industries. While the technology introduces groundbreaking features, such as altering voices and reimagining sounds, the company has no immediate plans to release it publicly.

Revealed on Monday, Fugatto—short for Foundational Generative Audio Transformer Opus 1—can produce music and sound effects based on text prompts. It also offers innovative capabilities like converting a piano melody into a human singing voice or modifying accents and emotional tones in spoken recordings.

“Generative AI will bring new capabilities to music, video games, and creators,” said Bryan Catanzaro, Nvidia’s Vice President of Applied Deep Learning Research. He compared its potential impact to that of computers and synthesizers on music production over the last half-century.

Fugatto enters a rapidly evolving space in generative audio and video technologies, competing with platforms like Runway and Meta. Nvidia trained the model using open-source data but is holding off on public release due to risks such as misinformation and copyright infringement.

The debut of generative AI models has sparked intense debate, particularly in entertainment, where copyright issues and voice imitation controversies—such as Scarlett Johansson’s recent dispute with OpenAI—have raised concerns. Nvidia’s cautious stance mirrors industry giants like OpenAI and Meta, which have also withheld similar technologies from public access.

As a leader in AI hardware and software, Nvidia’s careful approach with Fugatto highlights growing ethical and legal concerns surrounding generative AI, emphasizing the need for responsibility in its deployment.