Say hello to JEN-1, an AI model that can create music from text prompt, check details

Futureverse, a prominent AI company with diverse expertise in AI and Metverse technology, has unveiled its upcoming creation, JEM-1, a groundbreaking text-to-music converter. Jen-1 intends to overcome the shortcomings seen in current music generators such as Google's MusicLM, presenting superior audio quality and more attractive musical compositions than what is presently accessible in the industry.

Due to the sophisticated complexity of musical compositions and the need for a high sample rate, creating music from text has long been a difficult task. JEN-1 from Futureverse, on the other hand, overcomes these challenges by just sampling and letting the users type the music they want to generate into a prompt.

Features of JEN-1

One of the most remarkable features of JEN-1 is its impressive computational efficiency, which enables real-time music generation. This groundbreaking capability opens up an array of exciting opportunities for music production, live performances, and even virtual reality experiences.

The JEN-1 system uses a specialised autoencoder and diffusion model to generate accurate stereo audio at an astonishingly high sampling rate of 48kHz. The model successfully avoids the frequent quality degradation experienced during audio feature transfer. It is very adaptable because of its training, which covers a variety of tasks including music synthesis, music sequence continuation, and the addition of missing pieces.

JEN-1 triumphs over competitors

Futureverse subjected JEN-1 to a comprehensive evaluation, comparing it with leading models such as Google's MusicLM and Meta's MusicGen. The assessment utilised quantitative and qualitative measures, including the FAD (Fidelity-Awareness-Disentanglement) and CLAP (Continuity-and-Local-Anomaly-Penalties) scores, as well as human evaluations of music quality and alignment.

In addition to receiving top ratings from human judges, it received the highest FAD and CLAP scores. With just 22.6 percent of MusicGen's parameters and 57.7 percent of Noise2Music's, JEN-1's computational efficiency was further demonstrated.

An important step for AI in music

JEN-1 is a ground-breaking breakthrough that has the potential to change the landscape of music creation since it is the first model to excel in both quantitative and qualitative measurements in addition to its computational efficiency.

In conclusion, Futureverse's release of JEN-1 establishes a new benchmark in the combination of AI and music, making it possible to create high-quality music from text in real-time for the first time.