🔍 Read the full analysis: How Open ASR Is Expanding To Egyptian And Nigerian English on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Hugging Face has added two private Liva AI evaluation splits for Egyptian and Nigerian English to its Open ASR Leaderboard. The spontaneous conversational recordings total 2.89 hours; the announcement does not give speaker-composition details or report how the additions affect model rankings.
Hugging Face has added two private English speech-test splits from Liva AI to its Open ASR Leaderboard, expanding evaluations to include Egyptian and Nigerian English, as detailed in the original analysis. The spontaneous conversational recordings total 2.89 hours and are available to the leaderboard for assessment, not as ordinary public benchmark datasets.
The new splits are named Liva AI Conversational EG and Liva AI Conversational NG. Hugging Face says the Egyptian set contains 1.34 hours of speech and the Nigerian set contains 1.55 hours. Their transcripts are punctuated and cased, and retain disfluencies, according to the announcement.
Hugging Face says private evaluation data can make assessments more trustworthy by reducing developers’ ability to tune systems directly against test examples that are publicly available. It has also added two ways to compare results on the leaderboard’s Private tab: US versus non-US accents, and Inner Circle versus Outer and Expanding Circles.
The non-US group listed in the announcement includes the UK, Australia, Canada, India, Nigeria and Egypt. The circles comparison places the US, UK, Australia and Canada in the Inner Circle, and India, Nigeria and Egypt in the combined Outer and Expanding group. Hugging Face credits Hume AI with the idea for the leaderboard grouping and says the framework draws on Kachru’s Three Circles model.
Egyptian and Nigerian English join the Open ASR Leaderboard
Hugging Face added private Liva AI speech splits to broaden accent evaluation. The short recordings add coverage, while leaving questions about how representative the results are.
What was added
Both datasets contain spontaneous conversation with punctuated, cased transcripts that retain disfluencies.
Liva AI Conversational EG
A private evaluation split for Egyptian-accented English.
Liva AI Conversational NG
A private evaluation split for Nigerian-accented English.
New ways to compare accents
The leaderboard’s Private tab will show results across two groupings.
- Non-US includes the UK, Australia, Canada, India, Nigeria and Egypt.
- Inner Circle compared with Outer and Expanding Circles, based on Kachru’s framework.
- Hugging Face plans to separate the latter circles as more accents are added.
Why private evaluation matters
Keeping test examples private may make it harder to tune systems directly to known benchmark material.
Users can inspect recognition results for two additional English varieties.
Scores describe these specific short test sets, not all speakers in either country.
More Accent-Specific ASR Results
Speech recognition systems can perform differently across varieties of English. Adding Egyptian and Nigerian evaluation sets gives leaderboard users results for accents not included in its previously listed groupings. The results can be considered alongside overall scores, which do not show performance differences between speaker groups.
The two sets contain less than three hours of audio combined. The announcement does not establish how representative their speakers or conversations are. Scores describe performance on these specific test sets and do not provide a comprehensive measure of recognition quality for Egyptian or Nigerian English.
Private evaluation examples are not publicly available for developers to use when tuning models against the exact test recordings. The announcement provides no independent assessment of the data’s representativeness and no results showing whether the new splits change model scores or leaderboard rankings.
Products Worth Considering
Bose QuietComfort Ultra Earbuds (2nd Gen) – Wireless Noise‑Cancelling Earbuds with Mic, Immersive Audio, USB‑C Charging, Up to 6 Hours Batte
EcoFlow DELTA 3 Portable Power Station (1024Wh, 1800W)
Sonos Era 100 SL Compact Microphone-Free Speaker, White
As an affiliate, we earn on qualifying purchases.
Earlier Private Accent Evaluations
The Liva AI splits continue Hugging Face’s effort to add private speech tests to the Open ASR Leaderboard. The announcement says the platform previously added 11 private splits from Appen and DataOceanAI, covering scripted and conversational speech across five accents. It later added a private split for Indian-accented English in collaboration with Voice Arena.
The new accent comparisons organize results at a broader level as well as by specific accent. Hugging Face says it plans to separate the Outer and Expanding Circles into distinct groups as more accents become available. The announcement does not give a timeline or a required number of accents for making that change.
““We’re excited to announce a new collaboration with Liva AI to add even more private data with two new accents: Egyptian and Nigerian.””
— Hugging Face
What the Short Splits Reveal
The announcement does not say when the recordings were collected, how many speakers took part, or how they were selected. It also gives no breakdown by region, age or other characteristics that could help readers assess how the recordings reflect the range of English spoken in Egypt and Nigeria.
Hugging Face has not reported baseline scores or the effect of the new data on existing leaderboard rankings. It also does not say whether the recordings or more detailed dataset documentation will become public; the splits are described as private evaluation data. The announcement therefore leaves the datasets’ speaker composition and effect on rankings unspecified.
More Accents and Group Comparisons
Hugging Face says it plans to continue adding accents and languages to the leaderboard and intends to split the Outer and Expanding Circles into separate comparisons when more accents are available. No date or threshold for that change was given. For now, users can find the new comparisons on the leaderboard’s Private tab.
The announcement does not specify when details on speaker composition, collection methods or results by accent may be released, or when additional accents or updated leaderboard results may appear.
Key Questions
Which accents do the new Open ASR splits cover?
They cover Egyptian English and Nigerian English, in datasets named Liva AI Conversational EG and Liva AI Conversational NG.
How much speech is included?
The Egyptian split has 1.34 hours and the Nigerian split 1.55 hours, for 2.89 hours combined.
What kind of recordings are in the datasets?
Hugging Face describes them as spontaneous conversational audio. The transcripts are punctuated and cased, and include disfluencies.
Can developers download the evaluation recordings?
The splits are described as private evaluation data, not ordinary publicly available benchmark datasets. The announcement does not say whether access or documentation may change.
Do the additions show how well speech models work for all speakers in Egypt and Nigeria?
No. The announcement gives no speaker counts or selection details, and the datasets are short. Their scores reflect performance on these specific test sets, not a complete assessment of either country’s English varieties.
Primary source: Hugging Face · via ThorstenMeyerAI.com
