Replies: 6 comments
-
>>> reyxuan |
Beta Was this translation helpful? Give feedback.
-
>>> carlfm01 |
Beta Was this translation helpful? Give feedback.
-
>>> erogol |
Beta Was this translation helpful? Give feedback.
-
>>> carlfm01 |
Beta Was this translation helpful? Give feedback.
-
>>> erogol |
Beta Was this translation helpful? Give feedback.
-
>>> carlfm01 |
Beta Was this translation helpful? Give feedback.
-
>>> carlfm01
[October 10, 2019, 3:38am]
Hello, just to share the data that I'm currently using, it contains 50h
of reviewed speech and 50h of aligned speech but not reviewed. To review
I'm currently using a DeepSpeech model, where the transcription matches
the DS prediction I mark it as valid.
Most of the time we have a limited amount data to train (in this case
for Spanish), the idea of this dataset is to use it as base and try to
adapt it for a new a voice with way less data.
Using LJSpeech format!
Enjoy and please share any feedback.
drive.google.com
[.googledocs-onebox-logo
.g-drive-logo}
### tux-100h.zip
Google Drive file.
[This is an archived TTS discussion thread from discourse.mozilla.org/t/sharing-my-100h-of-single-speaker-spanish]
Beta Was this translation helpful? Give feedback.
All reactions