1 paper · 1 filter
Yao Shi, Ming Li
Currently, many multi-speaker speech synthesis and voice conversion systems address speaker variations with an embedding vector. Modeling it directly allows new voices outside of t…