Idea: pronunciation tips on recording/playback screen

Common Voice
#1

It’d be useful if we could long press or click on a word on the record/playback screen to get a link to Wiktionary’s pronunciation guide for that word. For example:

https://en.wiktionary.org/wiki/example#Pronunciation

This would make it easier to both add new sentences and review existing ones. Voice volunteer eyeballs on Wiktionary pages is likely to be a good thing for them too, and with enough coverage it could make an additional source of recording data. Everyone’s a winner!

Edit:

I’ve added this to GitHub as a feature request:

#2

Thanks for the feedback!

In your opinion, what’s the current problem you are observing that you feel that this solution would be useful to solve? How important is this problem and do you have any intuitions on why this is a problem?

Framing this as a problem statement would help us better understand it and see where to put it in our roadmap (if possible).

Thanks!

#3

In your opinion, what’s the current problem you are observing that you feel that this solution would be useful to solve?

I often see unfamiliar words and I’m not sure how to pronounce them. So when recording I either skip the sentence, or I give it my best shot. When reviewing I usually assume that the reader is correct.

How important is this problem

I don’t think it’s an enormous problem, but suspect that example pronunciations would help the quality and breadth of the data.

and do you have any intuitions on why this is a problem?

I think this is likely to result in sentences being added that have been read incorrectly. If I understand correctly, it’s very useful to have words that are commonly pronounced differently due to accent, but words that are mispronounced due to them being unfamiliar to the reader are, useless at best and damaging at worst.

1 Like
#4

When validating, I approve a clip with an obscure word if I look at that word and feel that the user has made a reasonable pronunciation attempt that is either close to a similarly-spelled word or obeys general English language rules. If it’s really obscure, phonetic pronunciation can be ok too IMO.

But I’m working to remove those kinds of sentences so hopefully unpronounceable words will be a far less frequent occurrence.

1 Like
#5

As @dabinat commented, we try to avoid extremely complex words to be listed, this should help with the problem you are commenting.

Can you clarify how often is “often”? Out of 100 clips, how many are you finding in this situation (estimated average from your observations).

Thanks!

#6

It’s actually got a lot better recently. I assume it’s because of the work to remove complex words.

What I’ll do is, I’ll make a note of where I’m up to, and count how many times I’ve pressed skip, then come back to you with an actual percentage figure in a week or so.

#7

I just recorded 55 and skipped 10. So about 15% for someone with my vocabulary and reading comprehension.

I’m not very well traveled but I am pretty technical, if that’s of relevance.

#8

I’ve found that sentence quality seems to vary because they’re randomly selected. So sometimes you’ll get a run of very straightforward, simple sentences and other times you’ll get a whole series of foreign place names or scientific terms one after the other.

1 Like