Primary competition visual

Mozilla Luganda Automatic Speech Recognition

Helping Uganda
$3 000 USD
Challenge completed almost 4 years ago
Automatic Speech Recognition
Natural Language Processing
179 joined
20 active
Starti
Oct 13, 21
Closei
Jan 16, 22
Reveali
Jan 16, 22
User avatar
AkashPB
The version of Common Voice Corpus to be used as a training data
Data · 23 Oct 2021, 14:31 · 2

Right now Common Voice Corpus version 7.0 is available. Kindly specify which version we need to use?

Since this is a 4-month long competition, and considering that Common Voice is updating their datasets at very frequent intervals, it may happen that in the next few months we may have a new version of Common Voice Corpus. Do we need to retrain our model with the updated one or do we need to proceed with the model which we will be training for version 7.0?

I am asking because training such models is a very time-consuming process and takes a lot of GPU time, it will be better if you specify which version of Common Voice Corpus to use so that we all are on a common page.

Thanks

Discussion 2 answers
User avatar
ZINDI

Only use versions equal to or less than version 7.0.

28 Oct 2021, 14:40
Upvotes 0
User avatar
AkashPB

Thanks @Zindi 🤗🤗