NVIDIA takes on Google, OpenAI, launches new AI model Fugatto that can create audio from texts

NVIDIA has launched Fugatto, a groundbreaking generative AI model capable of transforming text prompts into audio. Officially named the Foundational Generative Audio Transformer Opus 1, this experimental model has been designed to handle a variety of audio-related tasks, from creating music to modifying existing sounds.  

Described by NVIDIA as a “Swiss Army knife for sound,” Fugatto showcases advanced multilingual and multi-accent capabilities, thanks to its development by a global team of AI researchers.

Audio creation for professionals and beyond

Fugatto offers a range of applications for professionals in fields like music, language education, and game development. Music producers, for example, can use the AI to quickly draft song prototypes, experimenting with styles, voices, and instruments. Educators might find it useful for creating language-learning tools, tailoring audio to specific voices or accents. For game developers, Fugatto could dynamically adjust pre-recorded audio to align with gameplay changes, enhancing player immersion.

Beyond these direct applications, Fugatto can also handle complex tasks that combine instructions from its training data. For instance, it can create speech that conveys specific emotions, such as anger, in a chosen accent, or craft soundscapes that evolve over time, like a rainstorm moving across a landscape. These features demonstrate the AI’s adaptability and creative potential in audio generation.

More from Tech

How ChatGPT is becoming everyone’s BFF and why that’s dangerous America ready for self-driving cars, but it has a legal problem

A competitive space in generative AI for audio

While Fugatto’s capabilities are impressive, it’s entering a growing field of AI-driven audio tools. Meta has previously released an open-source AI kit for text-to-sound generation, and Google’s MusicLM allows users to create music from text prompts via its AI Test Kitchen platform.  

NVIDIA’s model, however, stands out with its emphasis on natural, human-like sound generation and its ability to modify existing audio files with precision.

No plans for public access

NVIDIA has not yet disclosed plans to make Fugatto publicly available. However, the model’s potential to revolutionise sound design is clear. From simplifying workflows for professionals to enabling more personalised audio experiences, Fugatto is another step forward in merging AI with creative expression.  

Whether or not it becomes accessible to the public, the model highlights NVIDIA’s ambition to redefine what’s possible in audio innovation.