Y
Hacker News
new
|
ask
|
show
|
jobs
by
arkobel
110 days ago
The lack of parallel accent data makes this fundamentally unsupervised. Curious if this leans more on latent disentanglement than direct supervision.