Fula - UNICEF WCARO NLP

🌍 General Information

ISO 639-3ful

ISO 639-1ff

French NameBiira, Boulbe, Domona, Dzemay, Eastern Fulani, Eastern Fulfulde, Fellata, Fillanci, Foulfoulde, Ful, Fula, Fula Fulbe, Fulani, Fulata, Fulatanchi, Fulbe, Fulfulde, Mbororo, Nagapelta, Palata, Paldena, Paldida, Pelta Hay, Peuhl

Glottocodefula1264

Language FamilyNiger–Congo

Writing SystemLatin Adlam Arabic ( Ajami )

L1 Speakers37 million

L2 Speakers2.7 million

Population37000000 (10⁶+)

Endangerment—

CountriesBenin, Burkina Faso, Cameroon, Central African Republic, Cote d'Ivoire, Democratic Republic of Congo, Gambia, Guinea, Guinea-Bissau, Nigeria, Mali, Mauritania, Niger, Senegal, Sierra Leone, Sudan, Togo

Also Known AsBiira, Boulbe, Domona, Dzemay, Eastern Fulani, Eastern Fulfulde, Fellata, Fillanci, Foulfoulde, Ful

WikipediaFula on Wikipedia

🗺️ Language Use

Data from CLEAR Global's Language Use Data Platform.

👥 Actors (4)

Digital Umuganda (Startup)

Rwandan AI and open data company building voice infrastructure for African languages. "Umuganda" is the Rwandan tradition of community service -- Digi...

Farmerline / Darli AI (Startup)

Ghanaian agritech company building technology solutions for smallholder farmers. Core platform Mergdata is a cloud-based farm management system used b...

LAfricaMobile (Startup)

Communication platform serving 300+ companies across Africa. Provides messaging (SMS, WhatsApp Business API, Voice), USSD, and AI-powered Text-to-Spee...

UBC Deep Learning & NLP Lab (University)

Research lab at the University of British Columbia led by Muhammad Abdul-Mageed, focused on deep representation learning and natural language socio-pr...

🤖 NLP & Tech Resources

Lanfrica

OLAC

CLDR

✓ AfroLID

✓ Google Translate: 2024

✓ M2M-100

✓ Mozilla Common Voice: ☑️

✓ MADLAD-400 Docs: 26 MAD-Docs

✓ MADLAD-400 Sentences: 5000 MAD-Sents

✓ FLEURS

✓ MMS-ASR

✓ MMS-TTS

✓ MMS-LID

✓ Omnilingual ASR

✓ WaxalNLP

Benchmarks

ASR (Speech-to-Text)

FLEURS-102 test

Model	CER	Source
facebook/mms-1b-fl102	13.8	reported

FLEURS-102 dev

Model	CER	Source
facebook/mms-1b-fl102	13.9	reported

PazaBench (aggregated)

Model	CER	WER	Source
facebook/omniASR-LLM-7B	14.0	44.0	reported
facebook/omniASR-LLM-7B-ZS	13.0	43.0	reported
facebook/omniASR-LLM-1B	15.0	47.0	reported
facebook/omniASR-LLM-3B	15.0	47.0	reported
facebook/omniASR-LLM-300M	16.0	47.0	reported
facebook/omniASR-CTC-7B	15.0	51.0	reported
facebook/omniASR-CTC-1B	15.0	52.0	reported
facebook/omniASR-CTC-3B	14.0	50.0	reported
facebook/omniASR-CTC-300M	16.0	56.0	reported
microsoft/paza-mms-1b-all	15.0	54.0	reported
facebook/mms-1b-all	15.0	53.0	reported
facebook/mms-1b-fl102	16.0	53.0	reported
microsoft/paza-Phi-4-multimodal-instruct	38.0	91.0	reported
nvidia/parakeet-tdt-0.6b-v3	38.0	90.0	reported
facebook/wav2vec2-conformer-rel-pos-large-960h-ft	32.0	95.0	reported
facebook/data2vec-audio-large-960h	36.0	94.0	reported
facebook/wav2vec2-conformer-rope-large-960h-ft	34.0	98.0	reported
facebook/hubert-large-ls960-ft	32.0	90.0	reported
facebook/hubert-xlarge-ls960-ft	33.0	91.0	reported
microsoft/paza-whisper-large-v3-turbo	46.0	92.0	reported
facebook/wav2vec2-large-960h-lv60-self	34.0	89.0	reported
facebook/data2vec-audio-base-960h	42.0	105.0	reported
facebook/wav2vec2-large-robust-ft-libri-960h	35.0	93.0	reported
nvidia/canary-qwen-2.5b	37.0	95.0	reported
facebook/wav2vec2-large-960h	52.0	100.0	reported
ibm/granite-speech-3.3-2b	45.0	99.0	reported
facebook/wav2vec2-base-960h	66.0	101.0	reported
openai/whisper-large-v3-turbo	56.0	90.0	reported
efficient-speech/lite-whisper-large-v3-turbo-acc	55.0	90.0	reported
distil-whisper/distil-large-v3	52.0	92.0	reported
ibm/granite-speech-3.3-8b	54.0	107.0	reported
openai/whisper-large	62.0	93.0	reported
distil-whisper/distil-medium.en	62.0	95.0	reported
openai/whisper-large-v3	71.0	97.0	reported
distil-whisper/distil-large-v2	66.0	95.0	reported
Qwen/Qwen2-Audio-7B-Instruct	53.0	98.0	reported
efficient-speech/lite-whisper-large-v3-acc	75.0	97.0	reported
openai/whisper-large-v2	75.0	97.0	reported
kyutai/stt-2.6b-en	71.0	122.0	reported
openai/whisper-medium.en	74.0	97.0	reported
openai/whisper-small.en	76.0	97.0	reported
efficient-speech/lite-whisper-large-v3	77.0	97.0	reported
openai/whisper-base.en	74.0	97.0	reported
efficient-speech/lite-whisper-large-v3-turbo	78.0	97.0	reported
efficient-speech/lite-whisper-large-v3-fast	81.0	98.0	reported
microsoft/Phi-4-multimodal-instruct	63.0	116.0	reported
openai/whisper-tiny.en	77.0	98.0	reported
nvidia/canary-1b-v2	48.0	106.0	reported
efficient-speech/lite-whisper-large-v3-turbo-fast	90.0	99.0	reported
usefulsensors/moonshine-base	101.0	123.0	reported
Qwen/Qwen2-Audio-7B	113.0	158.0	reported
usefulsensors/moonshine-tiny	163.0	248.0	reported

MT (Machine Translation)

GATONES

Model	direction	BLEU	chrF	Source
google/madlad400-3b-mt	ful→eng	7.8	26.8	reported
google/madlad400-3b-mt	eng→ful	3	21.4	reported
google/madlad400-7.2b-mt	ful→eng	8.3	26.5	reported
google/madlad400-7.2b-mt	eng→ful	3.3	21.3	reported
google/madlad400-10.7b-mt	ful→eng	9.4	29.2	reported
google/madlad400-10.7b-mt	eng→ful	3.7	22.9	reported
google/madlad400-8b-lm	ful→eng	1.3	13.6	reported
google/madlad400-8b-lm	eng→ful	0	3.3	reported

FLORES-200

Model	direction	BLEU	chrF	Source
google/madlad400-3b-mt	ful→eng	10.5	33.2	reported
google/madlad400-3b-mt	eng→ful	2.2	23.4	reported
google/madlad400-7.2b-mt	ful→eng	10.1	33.1	reported
google/madlad400-7.2b-mt	eng→ful	2.5	24	reported
google/madlad400-10.7b-mt	ful→eng	12.4	35.8	reported
google/madlad400-10.7b-mt	eng→ful	2.5	25.1	reported
google/madlad400-8b-lm	ful→eng	2.4	18.2	reported
google/madlad400-8b-lm	eng→ful	0	7.8	reported

NTREX

Model	direction	BLEU	chrF	Source
google/madlad400-3b-mt	ful→eng	14.4	35.8	reported
google/madlad400-3b-mt	eng→ful	4.8	27.6	reported
google/madlad400-7.2b-mt	ful→eng	14.2	36	reported
google/madlad400-7.2b-mt	eng→ful	5.3	28.1	reported
google/madlad400-10.7b-mt	ful→eng	15.8	37.5	reported
google/madlad400-10.7b-mt	eng→ful	5.1	28.8	reported
google/madlad400-8b-lm	ful→eng	2.4	18.8	reported
google/madlad400-8b-lm	eng→ful	0.1	4.3	reported

LLM (Large Language Models)

FLORES-200

Model	Log-perplexity	Source
goldfish-models/fuv_latn_full	165.36	reported

📝 Notes

Observations

Survey Focus 2 for Burkina Faso (2026 UNICEF WCAR Local Languages Survey, as "Fulfude")
Survey Focus 3 for Mali (2026 UNICEF WCAR Local Languages Survey, as "Peul/Fulfuldé/Pulaar")
Widespread cross-border language spoken across WCA (Burkina Faso, Cameroon, Guinea, Mali, Niger, Nigeria, Senegal...)

ASR Models (9)

Model	Size	Downloads	Likes
MahmoudAshraf/mms-300m-1130-forced-aligner	0.3B	4.0M	83
facebook/mms-1b-all	1.0B	2.0M	195
facebook/mms-1b-fl102	1.0B	8.7K	29
facebook/mms-1b-l1107	1.0B	280	12
Baamtu/whisper-pular-ggml-q5	—	39	0
bc7ec356/heep-universal	0.8B	31	0
onnx-community/mms-300m-1130-forced-aligner-ONNX	—	25	0
UBC-NLP/Simba-SLID-49	94.6M	15	0
romara-labs/mms-300m-1130-forced-aligner-ONNX	—	4	0

View on HuggingFace: ff: 9

TTS Models (25)

Model	Size	Downloads	Likes
k2-fsa/OmniVoice	—	606.0K	577
drbaph/OmniVoice-bf16	—	4.7K	13
splendor1811/omnivoice-vietnamese	0.6B	672	2
gluschenko/omnivoice-onnx	—	397	4
edwixx/OmniVoice	—	163	0
WANJIAX2197/OmniVoice	—	30	0
Bgeorge/OmniVoice	—	25	1
yash3120/OmniVoice	—	25	0
ex510/OmniVoice	—	19	0
razi08/OmniVoice	—	17	0
Manish993135/OmniVoice	—	16	0
riasatsk/OmniVoice	—	16	0
harsha7286/OmniVoice	—	15	0
lopho/OmniVoice	—	15	0
rnjema-unima/mms-tts-ful-baseline	36.3M	15	0
bilalRHCH/Arabic-text-to-speech	—	14	0
phucpx247/OmniVoice	—	14	0
gubernac/OmniVoice	—	13	1
preference1/OmniVoice	—	13	0
EBINSHAJI1999/OmniVoice	—	12	0

View on HuggingFace: ff: 24 | ful: 1

MT Models (32)

Model	Size	Downloads	Likes
google/madlad400-3b-mt	3B	329.0K	192
jbochi/madlad400-3b-mt	3B	76.4K	154
alirezamsh/small100	0.3B	3.9K	104
google/madlad400-10b-mt	11B	3.6K	131
google/madlad400-7b-mt	8B	2.7K	20
notjjustnumbers/madlad400-3b-mt-Q4_K_M-GGUF	3B	225	4
google/madlad400-7b-mt-bt	8B	201	7
jbochi/madlad400-7b-mt-bt	8B	191	8
santhosh/madlad400-3b-ct2	—	135	14
mtsdurica/madlad400-3b-mt-Q4_K_M-GGUF	3B	89	4
jbochi/madlad400-7b-mt	8B	79	7
jbochi/madlad400-10b-mt	11B	71	18
NikolayKozloff/madlad400-10b-mt-Q8_0-GGUF	11B	54	3
Nextcloud-AI/madlad400-3b-mt-ct2-int8	—	53	2
vonjack/opus-mt-mul-en-big	0.3B	39	4
CXDuncan/madlad400-3b-mt-optimized-quantized-onnx	—	31	4
morlayecis0003/nllb-200-pulaar-mairie	1B	30	0
CXDuncan/madlad400-3b-mt-optimized-onnx	—	20	0
Rohith04/ct2fast_m2m100_418M	—	20	3
Heng666/madlad400-3b-mt-ct2-int8	—	20	1

View on HuggingFace: ff: 30 | ful: 2

LLM Models (36)

Model	Size	Downloads	Likes
UBC-NLP/cheetah-base	—	1.1K	7
LLaMAX/LLaMAX3-8B-Alpaca	8B	370	32
google/madlad400-8b-lm	9B	183	11
LLaMAX/LLaMAX3-8B	8B	32	39
LLaMAX/LLaMAX2-7B	7B	21	5
LLaMAX/LLaMAX2-7B-Alpaca	7B	18	3
ZeppelinCorp/Charm_15	—	18	0
jbochi/madlad400-8b-lm	9B	14	8
tangledgroup/tangled-alpha-0.6-core	—	9	0
goldfish-models/fuv_latn_5mb	—	8	0
LLaMAX/LLaMAX2-7B-MetaMath	7B	7	0
tangledgroup/tangled-alpha-0.8-core	—	7	0
LLaMAX/LLaMAX2-7B-XNLI	—	6	1
tangledgroup/tangled-alpha-0.7-core	—	6	0
goldfish-models/fuv_latn_10mb	—	6	0
LLaMAX/LLaMAX2-7B-X-CSQA	—	5	0
tangledgroup/tangled-alpha-0.5-core	—	5	0
goldfish-models/ful_latn_full	—	2	0
goldfish-models/fuv_latn_full	—	2	0
goldfish-models/ful_latn_5mb	—	2	0

View on HuggingFace: ff: 30 | ful: 6

Datasets

ASR Datasets (34)

Dataset	Rows	Downloads	Likes
google/fleurs	—	46.3K	392
google/WaxalNLP	2.6M	12.3K	219
sarulab-speech/yodas2_sidon	1.5M	6.0K	60
claudefitz/WaxalNLP	2.6M	3.6K	0
google/xtreme_s	—	3.2K	70
BolajiJsAI/WaxalNLP	2.6M	2.5K	1
ngoloan/WaxalNLP	2.0M	2.1K	0
novelwolde36/WaxalNLP	2.5M	2.0K	1
Phsntom/WaxalNLP	2.6M	1.9K	0
Ephraimmm/WaxalNLPr	2.6M	1.8K	0
fiifinketia/WaxalNLP	2.5M	1.7K	1
zoelaenskii/zinny	2.6M	1.4K	1
0xzanuee/WaxalNLP	1.9M	1.4K	0
suryatmodulus/yodas2_sidon	—	1.1K	0
WueNLP/belebele-fleurs	65.1K	799	9
satwikguptha/xtreme_s	—	364	0
nanexxx/fleurs	—	313	0
bdallhrajh371/WaxalNLP	2.6M	250	0
YuRiVeRTi/V1Q	186.0K	201	3
cahya/fleurs	—	169	1

View on HuggingFace: ful: 24 | ff: 10

TTS Datasets (18)

Dataset	Rows	Downloads	Likes
google/WaxalNLP	2.6M	12.3K	219
sarulab-speech/yodas2_sidon	1.5M	6.0K	60
claudefitz/WaxalNLP	2.6M	3.6K	0
BolajiJsAI/WaxalNLP	2.6M	2.5K	1
ngoloan/WaxalNLP	2.0M	2.1K	0
novelwolde36/WaxalNLP	2.5M	2.0K	1
Phsntom/WaxalNLP	2.6M	1.9K	0
Ephraimmm/WaxalNLPr	2.6M	1.8K	0
fiifinketia/WaxalNLP	2.5M	1.7K	1
zoelaenskii/zinny	2.6M	1.4K	1
0xzanuee/WaxalNLP	1.9M	1.4K	0
suryatmodulus/yodas2_sidon	—	1.1K	0
WueNLP/belebele-fleurs	65.1K	799	9
bdallhrajh371/WaxalNLP	2.6M	250	0
YuRiVeRTi/V1Q	186.0K	201	3
lenoxaugo254/LenoxAI	—	10	1
Menoviar28/menov31	—	9	0
dijihax/Dataset	—	8	2

View on HuggingFace: ful: 11 | ff: 7

Translation Datasets (18)

Dataset	Rows	Downloads	Likes
google/smol	798.0K	2.8K	94
gsarti/flores_101	207.0K	1.6K	32
ayymen/Pontoon-Translations	3.6M	1.1K	18
Helsinki-NLP/opus_ubuntu	37.4K	354	4
YuRiVeRTi/V1Q	186.0K	201	3
slone/nllb-200-10M-sample	10.0M	168	13
severo/flores_101	207.0K	124	2
Helsinki-NLP/qed_amara	—	114	5
cis-lmu/GlotStoryBook	80.1K	67	7
FrancophonIA/gatitos	—	61	4
freococo/quran_multilingual_parallel	6.2K	59	4
kargaranamir/GlotStoryBook-backup	80.1K	21	0
jeffbrian/flores_101	—	19	0
FrancophonIA/Lexique_panafricain_de_la_femme_et_du_developpement	—	16	0
Bretagne/PolyWrite_clean	155	11	0
lenoxaugo254/LenoxAI	—	10	1
Menoviar28/menov31	—	9	0
dijihax/Dataset	—	8	2

View on HuggingFace: ff: 14 | ful: 4