Common Voice 2024 Roadmap Update: Video

Hello all, I just wanted to share a video from the Common Voice team describing a bit more about what we’re all working on as part of the 2024 roadmap


There are many things, that I can criticize, but I want to start from something, that I liked. I really wait CVSS, because it’s something incredible that can be really useful and can help to make project not so much boring.

Here I’m starting to criticize. It was a good decision to make sentence collector built-in in the main page of CV, and I’m glad it helped to increase the number of contributions, but this collector works not so well like other version in separate site, because now I can’t make a list of many sentences from the same source in one page. Also, text in the field is too big in my opinion, but this is not so important. So I think you need to return this feature (one sentence = one line) or at least add a button “Save my source”. It wastes too much time, because now I need to add them one-by-one and every time I should add the same source, and this is really slow and so annoying.

Go on. Domains. This is a something well, but at first glance only. I’m really sorry, but this is useful only when people know what does it mean and how to use them. And in this case, I don’t sure it has any sense, because CV has no guides to that, and I don’t sure if someone is writing them. And if no one is doing that, then it will be almost unuseful like language/accent tags in user profiles now. It’s impossible to work with data when everyone uses a personal way of writing/using the same things. “Country: Belarus”/“Country of birth: Belarus”/“Motherland: Belarus”/“I was born and lived in Belarus” etc. Same information, but it’s written in the different ways, and it’s impossible to work with it. When researcher/someone else wants to use the data, he doesn’t want to waste time to stick together this information. And the same problem will be with domains. Without any guides, everyone understands them in his personal way. For example, what does mean “Language Fundamentals (e.g. Digits, Letters, Money)”? It has examples, but they don’t help without an explanation. So I think domains can be useful, but if we have guides for them only. And now it’s just a waste of strength.

Next. Bugs and strange things on CV platform. Almost last 1–2 years CV had a bug, when after some sessions of clip listening, you have got an error, that can be almost impossible to fix on the client side. This was the main reason why I have stopped to contribute to CV then. Also, where is a button “Skip” in the mobile interface now? Why I need to have an account for skipping page with congrats, when I listening clips? How does report system work, because after I reported one sentence with grammatical mistakes, I can see it again, but with other audio clips. And this is not clear where I can read about it. Then I feel that it has no sense to report them. Guidelines are really short and not so clear also. So one of the biggest problem of CV, that the command should try to decide, is improving of features, that you already have, to make CV more user-friendly and writing of more high-quality guidelines.

Also, it can be good to make this project not so boring. Just “Dashboard” and “Goals” are not enough for that. Maybe, to increase number of tasks and add more activities. Spontaneous speech is great, and I’ll repeat, I really wait it, but you can add also possibility to categorize why audio clips are bad (misread, repeating of some word, there’s no speech, there’s speech of someone else etc). This is more difficult, but it’s possible.

I think that’s all my key points.