Jen’s Research &
Audio Demos
Jen’s AI research team consists of the world’s leading AI PhDs and scientists who collectively have over 30 published artificial intelligence research papers in prestigious journals such as T-PAMI, TKDE, TIP, SIGIR, AAAI, ICRA, NeurIPS, ICCV, CVPR, and ACM MM.
The initial JEN-1 research paper outlines our approach to text-guided universal music generation with omnidirectional diffusion models.
Doctrine Image
Our JEN-1 Composer research paper details our unified framework for high-fidelity multi-track music generation.
Doctrine Image
Detailing our JEN-1 DreamStyler research paper on customized musical concept learning via pivotal parameters tuning.
Doctrine Image
JEN-1 DreamStyler Demos
Jen’s upcoming StyleFilter™ technology, outlined in our DreamStyler research paper, allows users to upload a unique instrument, genre, or track to create a StyleFilter™ that styles a new masterpiece with the desired sound.
D.001
Genre StyleFilter™️
The Genre StyleFilterTM showcases how Jen learns from a training track in a particular genre and then allows a user to input a prompt to direct the style of that style on a new masterpiece.
Earthy tones, environmentally conscious, ukulele-infused, harmonic, breezy, easygoing, organic instrumentation
Fado Training Track
Generated Track
80s electronic track with melodic synthesizers, catchy beat and groovy bass
Fado Training Track
Generated Track
A grand orchestral arrangement with thunderous percussion, epic brass fanfares, and soaring strings, creating a cinematic atmosphere fit for a heroic battle
Klezmer Training Track
Generated Track
A dynamic blend of hip-hop and orchestral elements, with sweeping strings and brass
Klezmer Training Track
Generated Track
A dynamic blend of hip-hop and orchestral elements, with sweeping strings and brass
Flamenco Training Track
Generated Track
Rock with saturated guitars, a heavy bass line and crazy drum break and fills
Flamenco Training Track
Generated Track
Acoustic folk song to play during roadtrips: guitar choirs
Huayno Training Track
Generated Track
80s electronic track with melodic synthesizers, catchy beat and groovy bass
Huayno Training Track
Generated Track
Pop dance track with catchy melodies, tropical percussion, and upbeat rhythms, perfect for the beach
Huayno Training Track
Generated Track
Acoustic folk song to play during roadtrips: guitar choirs
Maqam Training Track
Generated Track
A grand orchestral arrangement with thunderous percussion, epic brass fanfares, and soaring strings, creating a cinematic atmosphere fit for a heroic battle
Muzak Training Track
Generated Track
A dynamic blend of hip-hop and orchestral elements, with sweeping strings and brass
Muzak Training Track
Generated Track
Earthy tones, environmentally conscious, ukulele-infused, harmonic, breezy, easygoing, organic instrumentation
Qawwali Training Track
Generated Track
Acoustic folk song to play during roadtrips: guitar choirs
Qawwali Training Track
Generated Track
Lofi slow bpm electro chill with organic samples
Urban Training Track
Generated Track
A dynamic blend of hip-hop and orchestral elements, with sweeping strings and brass
Urban Training Track
Generated Track
D.002
Instrument StyleFilter™️
The Instrument StyleFilterTM showcases how Jen learns from a particular instrument in a particular genre and then allows a user to input a prompt to direct the style of that style on a new masterpiece.
Pop dance track with catchy melodies, tropical percussion, and upbeat rhythms, perfect for the beach
Balalaika Training Track
Generated Track
Acoustic folk song to play during roadtrips
Kora Training Track
Generated Track
Pop dance track with catchy melodies, tropical percussion, and upbeat rhythms, perfect for the beach
Guqin Training Track
Generated Track
A punchy double-bass and a distorted guitar riff
Oud Training Track
Generated Track
A punchy double-bass and a distorted guitar riff
Panflute Training Track
Generated Track
A grand orchestral arrangement with thunderous percussion, epic brass fanfares, and soaring strings, creating a cinematic atmosphere fit for a heroic battle
Shamisen Training Track
Generated Track
Pop dance track with catchy melodies, tropical percussion, and upbeat rhythms, perfect for the beach
Sitar Training Track
Generated Track
JEN-1 Composer Demos
Jen’s Composer research focuses on enhancing the JEN-1 model by enabling the separation of generated audio into controllable stems. Jen describes a training strategy that progressively empowers the model, starting from single-track reconstruction and leading to the free composition of multi-track arrangements, unlocking limitless musical creativity. Users can use Jen’s Composer to refine and curate AI-generated tracks iteratively, following a more traditional human workflow.
C.001
Free conditional generation on text prompt
These demos outline four stems (drum, bass, melody, additional instrument) generated from an input prompt, along with a composition that combines the stems together.
Craft a track with funk, acoustic group genre, funny, smooth mood, city streets, big city, jazz guitar, acoustic guitar
Drum (generated)
Bass (generated)
Instrument (generated)
Melody (generated)
Mixed
Craft a track with funk, acoustic group genre, funny, smooth mood, city streets, big city, jazz guitar, acoustic guitar
Drum (generated)
Bass (generated)
Instrument (generated)
Melody (generated)
Mixed
C.002
Conditional generation on one stem and text prompt
This demo showcases the upload of one human input, then Jen’s completion of three additional instruments to complete the track.
Craft a track with funk, acoustic group genre, funny, smooth mood, city streets, big city, jazz guitar, acoustic guitar
Instrument (human input)
Drum (generated)
Bass (generated)
Melody (generated)
Mixed
Craft a track with funk, acoustic group genre, funny, smooth mood, city streets, big city, jazz guitar, acoustic guitar
Instrument (human input)
Drum (generated)
Bass (generated)
Melody (generated)
Mixed
C.003
Conditional generation on two stems and text prompt
This demo showcases the upload of two human inputs, then Jen’s completion of three additional instruments to complete the track.
Craft a track with funk, acoustic group genre, funny, smooth mood, city streets, big city, jazz guitar, acoustic guitar
Drum (human input)
Instrument (human input)
Bass (generated)
Instrument (generated)
Mixed
Craft a track with funk, acoustic group genre, funny, smooth mood, city streets, big city, jazz guitar, acoustic guitar
Drums (human input)
Instrument (human input)
Bass (generated)
Melody (generated)
Mixed
C.004
Conditional generation on three stems and text prompt
This demo showcases the upload of three human inputs, then Jen’s completion of three additional instruments to complete the track.
Craft a track with funk, acoustic group genre, funny, smooth mood, city streets, big city, jazz guitar, acoustic guitar
Drum (human input)
Bass (human input)
Instrument (human input)
Melody (generated)
Mixed
Craft a track with funk, acoustic group genre, funny, smooth mood, city streets, big city, jazz guitar, acoustic guitar
Drum (human input)
Instrument (human input)
Melody (human input)
Bass (generated)
Mixed
JEN-1 Omnidirectional Diffusion Models Demos
JEN-1’s unprecedented universal high-fidelity model for text-to-music demonstrates higher-quality outputs than the state-of-the-art baselines, such as MusicLM and MusicGen, previously released by Google and Facebook.
R.001
Comparison between music generative models
In the following, we compare the output of JEN-1 to the outputs of other state-of-the-art music generative models (MusicGen, MusicLM, Riffusion and Mousai).
A punchy double-bass and a distorted guitar riff
JEN-1
MusicGen
MusicLM
Riffusion
Mousai
Lofi slow bpm electro chill with organic samples
JEN-1
MusicGen
MusicLM
Riffusion
Mousai
Smooth jazz, with a saxophone solo, piano chords, and snare full drums
JEN-1
MusicGen
MusicLM
Riffusion
Mousai
A grand orchestral arrangement with thunderous percussion, epic brass fanfares, and soaring strings, creating a cinematic atmosphere fit for a heroic battle
JEN-1
MusicGen
MusicLM
Riffusion
Mousai
A dynamic blend of hip-hop and orchestral elements, with sweeping strings and brass
JEN-1
MusicGen
MusicLM
Riffusion
Mousai
Pop dance track with catchy melodies, tropical percussion, and upbeat rhythms, perfect for the beach
JEN-1
MusicGen
MusicLM
Riffusion
Mousai
Classic reggae track with an electronic guitar solo
JEN-1
MusicGen
MusicLM
Riffusion
Mousai
Earthy tones, environmentally conscious, ukulele-infused, harmonic, breezy, easygoing, organic instrumentation, gentle grooves
JEN-1
MusicGen
MusicLM
Riffusion
Mousai
Violins and synths that inspire awe at the finiteness of life and the universe
JEN-1
MusicGen
MusicLM
Riffusion
Mousai
80s electronic track with melodic synthesizers, catchy beat and groovy bass
JEN-1
MusicGen
MusicLM
Riffusion
Mousai
A piano and cello duet playing a sad chambers music
JEN-1
MusicGen
MusicLM
Riffusion
Mousai
A light and cheerly EDM track, with syncopated drums, aery pads, and strong emotions
JEN-1
MusicGen
MusicLM
Riffusion
Mousai
Acoustic folk song to play during roadtrips: guitar flute choirs
JEN-1
MusicGen
MusicLM
Riffusion
Mousai
Rock with saturated guitars, a heavy bass line and crazy drum break and fills
JEN-1
MusicGen
MusicLM
Riffusion
Mousai
R.002
Representation of key musical attributes.
In the following, we compare the output of JEN-1 to the outputs of other state-of-the-art music generative models (MusicGen, MusicLM, Riffusion and Mousai).
Genre
Moody, melancholy medium-tempo standard jazz song that is good for late-night listening.
JEN-1
Moody, melancholy medium-tempo fusion jazz song that is good for late-night listening
JEN-1
Moody, melancholy medium-tempo blues that is good for late-night listening.
JEN-1
Instrument
A romantic love song played by a band with a lead piano
JEN-1
A romantic love song played by a band with a lead saxophone
JEN-1
Tempo
Slow-paced progressive rock instrumental video game music with electric guitars, keyboards, bass and drums
JEN-1
Rapidly-paced progressive rock instrumental video game music with electric guitars, keyboards, bass and drums
JEN-1
Mood
Funky, bright and happy song with smooth, soulful male vocals
JEN-1
Romantic song with smooth, soulful male vocals
JEN-1
Era
Rock music from the 70s, with bass, drum, guitar
JEN-1
Rock music from the 80s, with bass, drum, guitar
JEN-1
Rock music from the 90s, with bass, drum, guitar
JEN-1
Rock music from the 2000s, with bass, drum, guitar
JEN-1
R.003
Zero-shot ability
It is possible to generate music based on creative prompts.
A funky hip hop song played with Scottish bagpipe
JEN-1
Acoustic guitars are playing heavy metal riffs
JEN-1
A flute plays a salsa song
JEN-1
Standard jazz song sung in acapella
JEN-1
Cartoon theme song sung by cats
JEN-1
R.004
Diversity
A dynamic blend of hip-hop and orchestral elements, with sweeping strings and brass, evoking the vibrant energy of the city
Variation 1
Variation 2
Variation 3
Variation 4
Variation 5
Smooth jazz, with a saxophone solo, piano chords, and snare full drums
Variation 1
Variation 2
Variation 3
Variation 4
Variation 5
R.005
Music inpainting (cut and edit)
By masking part of the tracks we can generate music that respects the text prompt, while following the remainder of the track - the following samples have masking applied at 2-5 seconds.
Intense, drama, climax, avengers, dark knight, man of steel
Masked
Inpainted
Motivational, driving, anthemic, uplifting, positive, underscore
Masked
Inpainted
Sad, emotional, dreamy, reflective, ambient, pensive
Masked
Inpainted
Upbeat, optimistic, joyous, cheerful, light, happy
Masked
Inpainted
R.006
Music Continuation
We can generate coherent music continuations that respect the text prompt - the following samples have masking applied during the second half of the track (5-10 seconds).
Intense, drama, climax, avengers, dark knight, man of steel
Masked
Inpainted
Motivational, driving, anthemic, uplifting, positive, underscore
Masked
Inpainted
Sad, emotional, dreamy, reflective, ambient, pensive
Masked
Inpainted
Upbeat, optimistic, joyous, cheerful, light, happy
Masked
Inpainted