MAGNeT - a facebook Collection

facebook 's Collections

Seamless Communication

MAGNeT

XLSR

XLS-R

Robust Wav2Vec 2.0

HuBERT

Fairseq S^2 TTS

Dinov2

MusicGen Stereo

MAGNeT

updated Apr 4

Masked Audio Generation using a Single Non-Autoregressive Transformer

Masked Audio Generation using a Single Non-Autoregressive Transformer

Paper • 2401.04577 • Published Jan 9 • 37
facebook/magnet-small-10secs

Text-to-Audio • Updated Jan 16 • 16

Note 300M model, text to music, generates 10-second samples.
facebook/magnet-medium-10secs

Text-to-Audio • Updated Jan 16 • 6

Note 1.5B model, text to music, generates 10-second samples.
facebook/magnet-small-30secs

Text-to-Audio • Updated Jan 16 • 7

Note 300M model, text to music, generates 30-second samples.
facebook/magnet-medium-30secs

Text-to-Audio • Updated Jan 16 • 30

Note 1.5B model, text to music, generates 30-second samples.
facebook/audio-magnet-small

Text-to-Audio • Updated Jan 16 • 8

Note 300M model, text to sound-effect.
facebook/audio-magnet-medium

Text-to-Audio • Updated Jan 16 • 26

Note 1.5B model, text to sound-effect.
facebook/hybrid-magnet-small

Text-to-Audio • Updated Feb 1 • 3
facebook/hybrid-magnet-medium

Text-to-Audio • Updated Feb 1 • 5