Hugging Face adds Egyptian and Nigerian English to Open ASR Leaderboard
Overview
Hugging Face adds Egyptian and Nigerian English evaluation data to its Open ASR Leaderboard, which ranks speech recognition models.
The data is private and comes through a collaboration with Liva AI. The Egyptian split has 1.34 hours of audio and the Nigerian split has 1.55 hours, both conversational and spontaneous, with punctuated, cased transcription.
The leaderboard will now report results separately for US versus non-US accents and by Kachru's Three Circles, a framework that groups English speakers by their relationship to the language.
Written by AI from the articles below · updated Nov 6, 5:28 AM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- Hugging Face BlogOfficialHugging Face adds Egyptian and Nigerian English to Open ASR Leaderboard
AIHugging Face adds private Egyptian and Nigerian English speech data to its Open ASR Leaderboard through a collaboration with Liva AI. The new Egyptian split has 1.34 hours of audio and the Nigerian split has 1.55 hours, both conversational and spontaneous with punctuated, cased transcription. The leaderboard will now break results down into US versus non-US accents and by Kachru's Three Circles.
Heat Trend
Not enough continuous observations to show a trend yet.