This is a great idea. A lot of us are constantly looking at something and most times there’s at least 2 people looking into the same thing, so it would be great for these conversations.
Mixed feelings. I’ve got to be brief as am on a coffee break from work!
I can definitely see the advantages for certain discussions. I’ve occasionally dropped in to the general Mozilla ML matrix channel but mostly out of interest rather than for a problem to solve - it seems a decent platform.
One aspect i find somewhat frustrating here and in GitHub Issues (and I sense I’m not alone ) is with people who put in zero effort around their question - I wonder what the move to a Matrix channel would do with them?
Does it give them a mainline straight into our consciousness via notifications (!) whereby those genuinely interested in the project get directly bothered by people who don’t make an effort or don’t get how OS software works (ie with demands etc) That’s probably me exaggerating it a little but they are quite annoying people!!
I don’t suggest Mozilla Matrix but https://app.element.io/ where you can just enable notifications for certain users who are interesting for you. It is also free (AFAIK) You can also move your conversation to a private chat for particular topics. I am also open to alternatives.
I suggest this because I generally see people try to communicate actively about different projects under irrelevant topics. This would enable real contributors to communicate on what matters. Also it makes easier for people working on similar things to find each other as @georroussos underlined.
In order to prevent noise, we can be explicit about the use of the chat, being only for development and project-specific purposes.
I think it is a great idea especially with the latest changes because it would definitely promote an active maintaining (which otherwise may not be there), since a lot of us think Mozilla TTS is now the most robust implementation from many sides, so it makes sense to keep building on this one whenever the state of the art changes. Since it is open source it can still be maintained and improved upon even if officially TTS is not the focus anymore. But most importantly, I think it is a good approach, especially for things we may stumble upon while trying out different stuff and discussions that may clarify something for someone. Additionally, if someone else is actively working on the same thing as another person, a small nugget of knowledge that one of these people discovers may go a long way for the project in general; and these discussions would promote this. I also agree with this being for people that have already spent a lot of time on the project (at least initially), because this way it just benefits everyone. Someone that has spent 100 hours on it is in a much better position to recognise shortcomings or promising approaches.
Generally I personally love the idea because as I am becoming more comfortable with the engine I’d like to contribute more things than just guidance for new people and these discussions would be a great place to start.
That would be a good starting point but would exclude non developers / non contributors to github like me. I am just testing, comparing and trying to integrate a TTS backend as a best compromise between quality and inference speed which could be valuable too.
Let us experiment with the idea and see how it works then we can setup another channel for the rest of the community. I am not even sure if using Element is good enough. So it is more of an experimental phase.
To be clear, we don’t have the intention to exclude anyone who is a part of our community in one way or another. Just want to emphasize that it is just an experiment.