search

vitrioil / Speech-Separation

Final Year Project for Speech Separation
MIT License
90 stars 7 forks source link

Catalyst #1

Closed Scitator closed 4 years ago

Scitator commented 4 years ago

Hi, Really nice project! Would you like to tell more about your project?

Please, join our slack, https://join.slack.com/t/catalyst-team-core/shared_invite/zt-d9miirnn-z86oKDzFMKlMG4fgFdZafw

vitrioil commented 4 years ago

Hey,

Our project is basically separating speech of individual speakers from a video using both audio and visual features. So, if a user selects a person in a video, the network should be able to separate out the individual's speech. This isn't our research, just an implementation of the paper Looking to Listen at a Cocktail Party and we used Pytorch+Catalyst for it and I have to say I was really bored to write the loops again and again so we chose Catalyst. I actually found out about catalyst from Kaggle. Really nice framework!

Thank you.