Where is Stable Audio 2.0 ? Why keep it or hide it ? Share it !
#20
by
Usedaniel
- opened
People need your true breakthrough model 'Stable Audio 2.0', why hide it or keep it ? Share it !
Usedaniel
changed discussion title from
Where is Stable Audio ? Why Hide it ?
to Where is Stable Audio ? Why keep it or hide it ? Share it !
Usedaniel
changed discussion title from
Where is Stable Audio ? Why keep it or hide it ? Share it !
to Where is Stable Audio 2.0 ? Why keep it or hide it ? Share it !
Stable Audio Open 1.0 is the same architecture as Stable Audio 2.0, except with a T5 text encoder instead of CLAP, and trained on a different dataset. The models trained on AudioSparx data (Stable Audio 1.0, 1.1, and 2.0) will not have open weights releases due to the data agreement with AudioSparx.
Fauno15
changed discussion status to
closed
Thank you Fauno15 for your detailed reply. I really hope that Stability AI will continue to serve its mission, releasing open weights, and so that investors will continue funding your hard and highly valuable work, for us and for humanity. Keep the good work !!!
aren't they training a different version with non audiosparx data?