AI4D Baamtu Datamation - Automatic Speech Recognition in WOLOF
$2,000 USD
Can you create an automatic speech recognition model for Wolof for use in public transport?
253 data scientists enrolled, 34 on the leaderboard
TransportationAutomatic Speech RecognitionNLPAudio
Senegal
12 February—23 May
Ends in 13 days
What are you guys doing to improve your score?
published 26 Apr 2021, 04:22

I am using wav2vec pretrained on 3/4 of the training data set. Did anyone added noise or removed silence or removed noise?

hello @kingabzpro, 😂😅 obviously almost nobody will answer your question(s) since they don't want to share their approaches until the competition ends.

adding noise and other stuff like that are what people experiment to come up with good scores so tbh you should try and see because even after competition ends, only some people will share their ideas.

Good luck !

I am from Kaggle, I think over there everyone shares ideas and helps each other. I have one more suggestion, there is much noise-filled audio in the training set, if you remove them from both training CSV and audio file we can improve the model.

you know what I want to give you another thing. Training on more than 4400 samples decreases the performance. If anyone can improve their score, I will be happy.

Yeah buddy you're really right! sharing and community habits are things we are working to improve here. but we are far from it today oh.