MODELHugging Face

5/5

ajesujoba/AfriHuBERT

Afar(aar)ASR/SLIDmHuBERT-147 (utter-project/mHuBERT-147)Encoder

View Source

Model Info

Language

Afar (aar)

Architecture

Encoder

Institution

Alabi, Jesujoba O.; Liu, Xuechen; Klakow, Dietrich; Yamagishi, Junichi

Last Updated

11 May 2026

Intended Use Cases

ASR/SLID for 1,226 African languages (incl. aar if in training data); -2.1% avg WER vs mHuBERT-147

License & Access

SPDX identifierCC

Licensed under CC. Review the source terms before use.

Provenance

Source URL

https://huggingface.co/ajesujoba/AfriHuBERT

Fetch Timestamp

11 May 2026

Submitter

Alabi, Jesujoba O.; Liu, Xuechen; Klakow, Dietrich; Yamagishi, Junichi

Full Record

20 of 25 fields stated

Every column the registry tracks. Values marked derivedwere filled by a rule or read from the source's text rather than stated in its metadata; the rule or the quoted sentence is shown under the value.

Registry columns (17/18)

iso code: aar
Model Name: ajesujoba/AfriHuBERT
Where is it hosted: HuggingFace
Link: https://huggingface.co/ajesujoba/AfriHuBERT
Year: 2024
Creator/Author(s): Alabi, Jesujoba O.; Liu, Xuechen; Klakow, Dietrich; Yamagishi, Junichi
Paper Reference: Alabi et al. (2024) — AfriHuBERT: A self-supervised speech representation model for African languages. arXiv:2409.20201
NLP Task: ASR/SLID
Architecture: Encoder
Base Model: mHuBERT-147 (utter-project/mHuBERT-147)
Fine-tuned?: No
License: CC
Open Access: Open
Has Model Card: Yes
SOTA Tasks: ASR/SLID for 1,226 African languages (incl. aar if in training data); -2.1% avg WER vs mHuBERT-147
Verified By: Not stated
Date Indexed: 2026-05-11
Comment: Self-supervised speech model covering 1,230 languages (1,226 African + Arabic/English/French/Portuguese). Afar likely included among 1,226 African languages but…

Readiness (LRS) (3/7)

model id: ajesujoba-afrihubert-aar
Model Type: Not stated
Training Data Sources: BibleTTS; Kallaama; MMS Ulab v2; NaijaVoices; NCHLT; Mozilla Common Voice; Nicolingua; VoxLingua107; Zambezi Voice (10K+ hours, 1,230 languages)
Benchmark Used: Not stated
Benchmark Score: Not stated
Benchmark Date: Not stated
Status: Activederivedupdated within 3 years

Readiness Score Breakdown

5/5

Data Coverage100%

Model Availability100%

Benchmarks100%

Documentation80%

Recency100%

Representativeness

Gender Balance

Regional Distribution

Dialect Coverage

Inclusivity

Relations Explorer link

Also Available In

This is the same multilingual corpus, sliced for other languages the platform indexes it under.

model

AfriHuBERT

HubertModel · ajesujoba

1,222 languagesAlumu-Tesu(aab)utter-project/mHuBERT-147

HUHugging Face

Restricted27 Oct 2026

model

AfriHuBERT

HubertModel · ajesujoba

1,222 languagesAfade(aal)utter-project/mHuBERT-147

HUHugging Face

Restricted27 Oct 2026

model

AfriHuBERT

HubertModel · ajesujoba

1,222 languagesAbidji(abi)utter-project/mHuBERT-147

HUHugging Face

Restricted27 Oct 2026

model

AfriHuBERT

HubertModel · ajesujoba

1,222 languagesAbua(abn)utter-project/mHuBERT-147

HUHugging Face

Restricted27 Oct 2026

model

AfriHuBERT

HubertModel · ajesujoba

1,222 languagesAbon(abo)utter-project/mHuBERT-147

HUHugging Face

Restricted27 Oct 2026

model

AfriHuBERT

HubertModel · ajesujoba

1,222 languagesAbron(abr)utter-project/mHuBERT-147

HUHugging Face

Restricted27 Oct 2026

model

AfriHuBERT

HubertModel · ajesujoba

1,222 languagesAbure(abu)utter-project/mHuBERT-147

HUHugging Face

Restricted27 Oct 2026

model

AfriHuBERT

HubertModel · ajesujoba

1,222 languagesKyode(acd)utter-project/mHuBERT-147

HUHugging Face

Restricted27 Oct 2026

model

AfriHuBERT

HubertModel · ajesujoba

1,222 languagesAcholi(ach)utter-project/mHuBERT-147

HUHugging Face

Restricted27 Oct 2026

model

AfriHuBERT

HubertModel · ajesujoba

1,222 languagesAcheron(acz)utter-project/mHuBERT-147

HUHugging Face

Restricted27 Oct 2026

model

AfriHuBERT

HubertModel · ajesujoba

1,222 languagesDangme(ada)utter-project/mHuBERT-147

HUHugging Face

Restricted27 Oct 2026

model

AfriHuBERT

HubertModel · ajesujoba

1,222 languagesDzodinka(add)utter-project/mHuBERT-147

HUHugging Face

Restricted27 Oct 2026

model

AfriHuBERT

HubertModel · ajesujoba

1,222 languagesAdele(ade)utter-project/mHuBERT-147

HUHugging Face

Restricted27 Oct 2026

model

AfriHuBERT

HubertModel · ajesujoba

1,222 languagesAdjoukrou(adj)utter-project/mHuBERT-147

HUHugging Face

Restricted27 Oct 2026

model

AfriHuBERT

HubertModel · ajesujoba

1,222 languagesAgotime(adq)utter-project/mHuBERT-147

HUHugging Face

Restricted27 Oct 2026

model

AfriHuBERT

HubertModel · ajesujoba

1,222 languagesTunisian Arabic(aeb)utter-project/mHuBERT-147

HUHugging Face

Restricted27 Oct 2026

model

AfriHuBERT

HubertModel · ajesujoba

1,222 languagesSaidi Arabic(aec)utter-project/mHuBERT-147

HUHugging Face

Restricted27 Oct 2026

model

AfriHuBERT

HubertModel · ajesujoba

1,222 languagesAmbele(ael)utter-project/mHuBERT-147

HUHugging Face

Restricted27 Oct 2026

model

AfriHuBERT

HubertModel · ajesujoba

1,222 languagesPutukwam(afe)utter-project/mHuBERT-147

HUHugging Face

Restricted27 Oct 2026

model

AfriHuBERT

HubertModel · ajesujoba

1,222 languagesAfo(afo)utter-project/mHuBERT-147

HUHugging Face

Restricted27 Oct 2026

Related Models

model

UBC-NLP/toucan-base

0.6B-parameter Afrocentric MT model. Supports Afar (aar) among 46 languages in 150+ pairs. Fine-tuned versions: toucan-1.2B and toucan-3.7B also available. T…

42 languagesAfar(aar)mT5MT

HUHugging Face

Restricted11 May 2026

model

UBC-NLP/toucan-1.2B

1.2B-parameter fine-tuned MT variant of Cheetah-1.2B. Significantly outperforms other MT models for African languages. Afar (aar) supported. Also available: …

42 languagesAfar(aar)mT5MT

HUHugging Face

Restricted11 May 2026

model

UBC-NLP/cheetah-1.2B

1.2B-parameter Afrocentric sequence-to-sequence pretrained LM covering 517 African languages. Afar (aar) included. Trained on 42G corpus from 50 of 54 Africa…

42 languagesAfar(aar)T5LM

HUHugging Face

Restricted11 May 2026