Despite the fact that artificial intelligence focuses on ChatGPT, Google Bard and other alternatives that focus on text and images, Meta just introduced MusicGen, your AI for creating and mixing music. It is completely free and is already available to users.
It is an artificial intelligence model capable of generating music from a textual description, with the ability to use an external audio track to guide the creationand is based on the comprehension model Transformer, created by Google.
This developed artificial intelligence was formed with over 20,000 hours of licensed musicfeaturing 10,000 high-quality tracks plus stock music repertoires Shutterstock and Pond5. He can be a great rival for MusicLM, the alternative that Google officially launched last May.
Create music from text
MusicGen is Meta’s new generative artificial intelligence, and its goal is create music from user-supplied text. This can be supplemented by an audio file from which the melody can be extracted to integrate it into the creation.
Although you can’t precisely control the orientation it will have in the generated audio. For the same description, there may be several different results since the AI performs the entire generation process each time it is requested.
Of course, this artificial intelligence it is only able to create a 12 second song snippet. This means that, at the moment, you cannot create whole songs, only small fragments. However, this AI has great possibilities.
Tests have been carried out with three versions of this model, the most basic has 300 million parameters, while the others have 1,500 and 3,300 million parameters, respectively. Theoretically, the larger the model, the better the quality of composition it offers, although according to The Decoder
How to use MusicGen
Mark Zuckerberg’s company AI has been released as open source and can be downloaded from GitHub, but you can also access a demo on the Hugging Face project portal. To test it, you just have to go to this portal, write a description in English and click on the Generate button. You can also condition the operation of the artificial intelligence by downloading an audio file from which you can extract the melody.
The estimated time it takes for the creative to appear varies depending on how many people tried to use this AI demo at the same time, but it may last more than 5 minutes. Note that you can’t control how the melody is used, but it can be generated multiple times to change the outcome.
In the tests we performed from EL ESPAÑOL – El Androide Libre, many requests ended with the AI displaying an error message due to the connection. On other occasions, once the time has elapsed, we have been able to listen to the 12 seconds that have been generated, and it must be said that the results are unpredictable, but they are very successful.
At the bottom of the page, Meta offers a series of examples to elaborate the descriptions. If you click on the name of the audio files, they will be added to the file box so that they can be used, for example.
You may be interested
Follow the topics that interest you