The applications and possibilities that all kinds of language models offer for content creation are actually quite extensive. Suffice it to say that even the common ChatGPT turns out to be an extremely useful tool for people who need help with the so-called creative block.
SunoAI, a tool for creating your own songs based on lyrics and appropriate prompts, is also quite popular among music fans. The creativity created in this tool sometimes seems to have no limits, which is actually very nice, especially for those who are not afraid of such experiments.
A new tool was created for such people, this time developed by NVIDIA. What is Fugatto? What possibilities does it have?
Fugatto, a tool for creating sounds
Fugatto (from English Foundational Generative Audio Transformer Opus 1) is an advanced GenAI model developed by NVIDIA engineers that allows you to manipulate sound using text commands. It can create music, replace voices, add effects and create completely unique sounds. Fugatto handles multiple tasks simultaneously and combines various instructions such as accent or emotion in the voice.
The scale with which this model was created is also impressive. We are talking about 2.5 billion parameters that are used by Fugatto for its needs. Of course, all this is powered by NVIDIA technologies, and the whole thing was created and developed by a team from different countries. Thanks to its innovative functions, Fugatto allows users to conduct artistic experiments on an unprecedented scale.
It is therefore not surprising that the engineers called their model a “Swiss army knife” for sound. The possibilities are enormous, because this technology was trained on a lot of data. It was available, among others: BBC Sound Library, which gives you access to a wealth of source material that is truly impressive.
Huge possibilities
But what can Fugatto be used for? Well, you can cite the example of music producers who will be able to create a “sketch” of a song based on prompts. They will also be able to conveniently add effects or try to adjust different styles, instruments or effects with just a few commands.
Our today’s hero will also be able to improve the overall quality of existing tracks or allow for the isolation of individual instruments. There are so many possibilities and they are not limited only to the music industry.
As has been established, Fugatto also operates on sounds, and this may allow game developers to adapt sounds to dynamic situations. This may translate into greater individuality for each player, who will hear different sounds or differently amplified dialogues depending on the situation.
The situation is similar in the case of marketing, where this model is able to adapt the sound and accent of the narrator to a specific region. This is a great simplification for all those who intend to create advertising campaigns in the future and reach recipients from all over the world.
When will Fugatto be available for general use?
According to the engineers behind Fugatto, we are entering a new era of music and sound creation, where AI will be our best assistant. Interestingly, this enthusiasm is shared by famous producer Ido Zmishlany.
In fact, in this case, the only barrier may seem to be our own creativity. I wonder how the NVIDIA model will cope with the multitude of interested users.
For now, however, NVIDIA has not revealed when we can expect Fugatto to be released and made available to a wider audience. For now, we can only wait and hope that it will happen sooner rather than later.
