Inject templates for ASR datasets #2565
Conversation
This reverts commit daf34ea.
|
Wait until #2567 is merged so we can benefit from the tagger :) |
datasets/common_voice/README.md
Outdated
| - zh-CN | ||
| - zh-HK | ||
| - zh-TW | ||
| - zh-Hans-CN | ||
| - zh-Hans-HK | ||
| - zh-Hant-TW |
There was a problem hiding this comment.
We should keep zh-CN, zh-HK and zh-TW since they are valid codes according to BCP-47
The Hans subtag means that the text uses the Simplified Chinese script
Therefore I'd suggest to add these three to our list of supported languages
There was a problem hiding this comment.
sounds good, will add them to datasets!
datasets/common_voice/README.md
Outdated
| @@ -43,11 +44,10 @@ languages: | |||
| - mt | |||
| - nl | |||
| - or | |||
| - pa-IN | |||
There was a problem hiding this comment.
we could add this language to our list of supported languages
datasets/common_voice/README.md
Outdated
| - rm-sursilv | ||
| - rm-vallader |
There was a problem hiding this comment.
these are valid BCP-47 codes according to wikipedia, we can add them to our list of languages
|
thanks for the feedback @lhoestq! i've added the new language codes and this PR should be ready for a merge :) |
This PR adds ASR templates for 5 of the most common speech datasets on the Hub, where "common" is defined by the number of models trained on them.
I also fixed a bunch of the tags in the READMEs 😎