Egyptian and Nigerian English added to Open ASR Leaderboard

The Open ASR Leaderboard has added two private evaluation splits for Egyptian and Nigerian English through a collaboration with Liva AI. The Egyptian split contains 1.34 hours of conversational speech, while the Nigerian split contains 1.55 hours. Both datasets consist of spontaneous conversations and provide punctuated, cased transcriptions that retain disfluencies.

The additions expand the leaderboard’s private evaluation data, which previously covered multiple accents through contributions from Appen, DataOceanAI and Voice Arena. The organizers said the effort is intended to improve accent representation in automatic speech recognition evaluations and increase trust by using private data and more providers.

The leaderboard will also offer a broader accent breakdown based on Kachru’s Three Circles. Results will be grouped into US and non-US accents, as well as Inner Circle accents—US, UK, Australia and Canada—versus Outer and Expanding Circle accents—India, Nigeria and Egypt. The organizers said those latter groups may be separated further as additional accents are added. Liva AI describes its data as coming from multimodal interactions with real users of consumer products.