Dear  all,

 

We would like to share with you about an exciting initiative called ML-SUPERB. The SUPERB team is organizing a multilingual speech self-supervised learning challenge. This year, they're introducing a new language track, which encourages participants to submit their spoken language resources (i.e., speech data with transcription).

 

In this track, participants will have the opportunity to test some common speech self-supervised models on their languages and submit related data to the challenge. This data will be used to evaluate the self-supervised models' submissions as a "hidden set." With their agreement, the data will also be included in future versions of the ML-SUPERB benchmark for further research.

For more information about the challenge, you may visit our website at https://multilingual.superbbenchmark.org/. Specifically, you can find details about the new language track at https://multilingual.superbbenchmark.org/challenge-asru2023/challenge_overview#Track2:-New-langauge-Track.

 

On Behalf of ML-SUPERB Organizers