Possible collaboration on CTXtxt2vec

[francislata]

Hi @cantabile-kwok, I’ve been chipping away on the unofficial implementation of the UniCATS paper here. Since the second part is out and it sounds like you’re working on the txt2vec portion of it, is there some possibility to collaborate on this? My unofficial repo contains some very basic dataset pre-processing and the different configs for establishing the contexts for each utterance.

Please do let me know. Thank you!

[cantabile-kwok]

I see, thanks so much for the interest in our work!

The tool you provided can be useful for open-sourcing. Recently I will work on the model part when I have time, and after that maybe I will take a look of how to integrate yours for better user experience. I will reach out to you when the process goes forward : )

[francislata]

Sounds good @cantabile-kwok. I’ll go through the current setup you have so that we have the same one.

Then feel free to get in touch with me if you have further updates. I’ll also take a look at the model part once I’ve combined the setup into my repo version, and we can consider the collaboration from there.

[cantabile-kwok]

@francislata Hi Francis, the model and training part of CTX-text2vec is released at https://github.com/cantabile-kwok/UniCATS-CTX-text2vec. Actually it's been a while, but I recently struggle to take time for the completion of inference scripts (including continuation, editing, and vocoding using CTX-vec2wav). You can check out the finished part, and if possible, would you like to help on the rest part, especially how to properly pipeline the text2vec and vec2wav models? Any PR would be greatly appreciated, or you can combine them in your repo and I will refer to yours in mine. 😄