Speech samples for the paper "Training Multi-Speaker Neural Text-to-Speech Systems using Speaker-Imbalanced Speech Corpora" which is submitted to INTERSPEECH 2019.
A pre-print version of this paper can be found at https://arxiv.org/abs/1904.00771
samples synthesized using WaveNet vocoder
1st sample
SD | MU | EN | CO | |
---|---|---|---|---|
XS01 | ► Play | ► Play | ► Play | ► Play |
XS02 | ► Play | ► Play | ► Play | ► Play |
S03 | ► Play | ► Play | ► Play | ► Play |
S04 | ► Play | ► Play | ► Play | ► Play |
S05 | ► Play | ► Play | ► Play | ► Play |
M06 | ► Play | ► Play | ► Play | ► Play |
M07 | ► Play | ► Play | ► Play | ► Play |
M08 | ► Play | ► Play | ► Play | ► Play |
L09 | ► Play | ► Play | ► Play | ► Play |
XL10 | ► Play | ► Play | ► Play | ► Play |
2nd sample
SD | MU | EN | CO | |
---|---|---|---|---|
XS01 | ► Play | ► Play | ► Play | ► Play |
XS02 | ► Play | ► Play | ► Play | ► Play |
S03 | ► Play | ► Play | ► Play | ► Play |
S04 | ► Play | ► Play | ► Play | ► Play |
S05 | ► Play | ► Play | ► Play | ► Play |
M06 | ► Play | ► Play | ► Play | ► Play |
M07 | ► Play | ► Play | ► Play | ► Play |
M08 | ► Play | ► Play | ► Play | ► Play |
L09 | ► Play | ► Play | ► Play | ► Play |
XL10 | ► Play | ► Play | ► Play | ► Play |
The synthetic speech samples were constructed using a Japanese multi-speaker speech database owned by KDDI Research