Futureverse has announced the release of the alpha version of Jena new model of generative artificial intelligence which enables create music from text prompts that the user gives. Available for anyone to use from their website, Jen is an AI model that can be used without having to worry about potential consequences resulting from copyright infringement. Has been trained with 40 catalogs fully licensed and verified against a database of 150 million songsensuring that creators can use the platform safely.
Some of Jen’s abilities that are available in her alpha release include: generate 10 or 45 second song fragments using text prompts. If you want to extend the song to a full length of 3:30 minutesyou can use the “continuation” and introduce new instructions for adding additional segments to the song. With a focus on “her commitment to transparency, compensation and copyright identification,” according to her press release, Jen has made much of her inner workings available to the public through research documents. They explain that the model uses “latent diffusion,” the same process used by Stable Diffusion, DALL-E and Image to create high-quality images.
Today, we couldn’t be more excited to premiere Jen (ALPHA). 🎵
Our ethically-trained generative AI music platform introduces a new standard for copyright compliance in text-to-music generation. Now, you can make music with your mind on https://t.co/iRqCeonM4N. pic.twitter.com/jQgkenMfou
— Jen (@jenmusicai) June 20, 2024
Jen also offers a verification for each work created. Users receive an indicator JENUINEa cryptographic hash registered on The Root Network blockchain, which certifies the authenticity of each composition.
Jen arrives months after meeting other text-to-music models such as Audiocraft of goal, DreamTrack from YouTube, Music LM from Google or Sunosomething that Futureverse motivates by having made sure to obtain all the necessary licenses, collaborating closely with major music companies and rights holders.
Its operation could not be simpler. The user enters the description of the music they want to generate, either detailing the style, a certain atmosphere or how each instrument should sound. It is also possible define them according to the style of an artist and the AI responds appropriately. What Jen doesn’t believe are voices.only musical instruments, but this ability will come in the future.
10-second fragments cost 1 credit and 30-second fragments cost 3 of the 300 that the user can consume and are replenished periodically. If you need more, Futureverse offers various rates to purchase them. You can also choose between the audio format, MP3 either WAVto download the created file, and if you prefer that the composition finish with a cut or progressively decreasing the volume.
Jen is in an alpha version, which means that it is software in an early stage of development and unpolished, although it seems to work normally, delivering exactly what it promises. The beta version will arrive next September and with it some capabilities will be expanded, offering longer duration in clips, Inpaintingconsisting of filling in missing sections in a piece of music, and StyleFilter. This latter feature will allow users to upload audio snippets of an instrument or track and modify its genre or timbre with a single click. It will also feature more licensed music for AI training and Futureverse will launch a beat market called R3CORDwhich will allow users to sell works created with Jen.