Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Amount of data to train own voice #165

Closed
Rachine opened this issue Apr 23, 2019 · 2 comments
Closed

Amount of data to train own voice #165

Rachine opened this issue Apr 23, 2019 · 2 comments

Comments

@Rachine
Copy link

Rachine commented Apr 23, 2019

Hey, I am trying to build a model with your version of Tacotron and my own data in US english.

As I am collecting data and formatting it, I am wondering what is the amount of data necessary to start getting some good results for the target voice? Does anyone know any empirical experiments so that I can set a target.

I have around 3/4h right now.

Thanks for the hard work on the repo
@erogol
Copy link
Contributor

erogol commented Apr 23, 2019

Amazon has some papers about the amount of the data needed for TTS. I'd say you can try to finetune one of the released models with your own data. That'd be the easiest way to go. You can also start from scratch but I'd guess the data is not sufficient. In general, my personal estimation is around 15 hours for something reasonable.

@erogol erogol closed this as completed Apr 23, 2019
@Rachine
Copy link
Author

Rachine commented Apr 24, 2019

Alright thank you very much! Will try that, and tell you if that improves it 👍

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@erogol @Rachine