Dyula - UNICEF WCARO NLP

🌍 General Information

ISO 639-3dyu

ISO 639-1—

French Namedioula

Glottocodedyul1238

Language FamilyNiger–Congo

Writing SystemN'Ko , Latin , Ajami

L1 Speakers2.6 million

L2 Speakers10 million

Population2500000 (10⁶+)

Endangermentsafe

CountriesBurkina Faso, Cote d'Ivoire, Gambia, Guinea, Mali, Senegal

Also Known AsDioula, Dioula Véhiculaire, Diula, Djula, Dyoula, Jula, Jula Kong, Julakan, Kong Jula, Tagboussikan

WikipediaDyula on Wikipedia

🗺️ Language Use

Data from CLEAR Global's Language Use Data Platform.

👥 Actors (2)

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...

ML Collective (Ngo)

Non-profit research lab providing research opportunities to independent and underrepresented researchers. Runs multiple African language NLP projects ...

🤖 NLP & Tech Resources

✓ AfroLID

✓ Google Translate: 2024

✓ NLLB-200

✓ Mozilla Common Voice

✓ MADLAD-400 Docs: 483 MAD-Docs

✓ MADLAD-400 Sentences: 19700 MAD-Sents

✓ Fineweb2: 💬

✓ CMU Wilderness

✓ MMS-ASR

✓ MMS-TTS

✓ MMS-LID

✓ Omnilingual ASR

✓ SimbaBench

🎙️ Common Voice

Total Hours0.5

Total Clips295

Validated211

Gender♂ 39.0% / ♀ 30.0%

Benchmarks

ASR (Speech-to-Text)

PazaBench (aggregated)

Model	CER	WER	Source
facebook/omniASR-LLM-7B	14.0	51.0	reported
facebook/omniASR-LLM-7B-ZS	10.0	40.0	reported
facebook/omniASR-LLM-1B	13.0	46.0	reported
facebook/omniASR-LLM-3B	13.0	50.0	reported
facebook/omniASR-LLM-300M	12.0	49.0	reported
facebook/omniASR-CTC-7B	15.0	58.0	reported
facebook/omniASR-CTC-1B	14.0	54.0	reported
facebook/omniASR-CTC-3B	16.0	63.0	reported
facebook/omniASR-CTC-300M	18.0	71.0	reported
microsoft/paza-mms-1b-all	13.0	49.0	reported
facebook/mms-1b-all	14.0	52.0	reported
facebook/mms-1b-fl102	38.0	100.0	reported
microsoft/paza-Phi-4-multimodal-instruct	40.0	98.0	reported
nvidia/parakeet-tdt-0.6b-v3	37.0	97.0	reported
facebook/wav2vec2-conformer-rel-pos-large-960h-ft	44.0	105.0	reported
facebook/data2vec-audio-large-960h	46.0	102.0	reported
facebook/wav2vec2-conformer-rope-large-960h-ft	46.0	105.0	reported
facebook/hubert-large-ls960-ft	43.0	100.0	reported
facebook/hubert-xlarge-ls960-ft	43.0	100.0	reported
microsoft/paza-whisper-large-v3-turbo	75.0	99.0	reported
facebook/wav2vec2-large-960h-lv60-self	44.0	100.0	reported
facebook/data2vec-audio-base-960h	52.0	110.0	reported
facebook/wav2vec2-large-robust-ft-libri-960h	45.0	102.0	reported
nvidia/canary-qwen-2.5b	50.0	109.0	reported
facebook/wav2vec2-large-960h	57.0	108.0	reported
ibm/granite-speech-3.3-2b	48.0	101.0	reported
facebook/wav2vec2-base-960h	60.0	107.0	reported
openai/whisper-large-v3-turbo	74.0	94.0	reported
efficient-speech/lite-whisper-large-v3-turbo-acc	60.0	95.0	reported
distil-whisper/distil-large-v3	71.0	96.0	reported
ibm/granite-speech-3.3-8b	49.0	103.0	reported
openai/whisper-large	52.0	94.0	reported
distil-whisper/distil-medium.en	63.0	98.0	reported
openai/whisper-large-v3	62.0	96.0	reported
distil-whisper/distil-large-v2	64.0	96.0	reported
Qwen/Qwen2-Audio-7B-Instruct	53.0	101.0	reported
efficient-speech/lite-whisper-large-v3-acc	61.0	96.0	reported
openai/whisper-large-v2	75.0	96.0	reported
kyutai/stt-2.6b-en	65.0	104.0	reported
openai/whisper-medium.en	78.0	96.0	reported
openai/whisper-small.en	78.0	96.0	reported
efficient-speech/lite-whisper-large-v3	68.0	97.0	reported
openai/whisper-base.en	79.0	97.0	reported
efficient-speech/lite-whisper-large-v3-turbo	76.0	97.0	reported
efficient-speech/lite-whisper-large-v3-fast	78.0	97.0	reported
microsoft/Phi-4-multimodal-instruct	56.0	106.0	reported
openai/whisper-tiny.en	117.0	98.0	reported
nvidia/canary-1b-v2	42.0	98.0	reported
efficient-speech/lite-whisper-large-v3-turbo-fast	86.0	99.0	reported
usefulsensors/moonshine-base	61.0	99.0	reported
Qwen/Qwen2-Audio-7B	47.0	97.0	reported
usefulsensors/moonshine-tiny	60.0	112.0	reported

SimbaBench (Common Voice 19)

Model	WER	CER	Source
facebook/mms-1b-all	66.59	16.45	reported
openai/whisper-large-v3	293.2	192.21	reported
openai/whisper-large-v3-turbo	108.37	44.4	reported
facebook/seamless-m4t-v2-large	150.85	101.24	reported
UBC-NLP/Simba-S	77.83	23.04	reported
UBC-NLP/Simba-M	67.86	22.17	reported
UBC-NLP/omniASR_LLM_7B_v2	37.94	9.03	reported
UBC-NLP/omniASR_LLM_1B_v2	50.29	12.47	reported
UBC-NLP/omniASR_LLM_300M_v2	45.41	11.76	reported
UBC-NLP/omniASR_CTC_7B_v2	55.73	15.82	reported
UBC-NLP/omniASR_CTC_3B_v2	55.74	13.46	reported
UBC-NLP/omniASR_CTC_1B_v2	60.49	14.89	reported

MT (Machine Translation)

FLORES+ 10-sample

Model	direction	SpBLEU	ChrF	Source
anthropic/claude-opus-4.6	dyu→eng	5.2	28.8	reported
anthropic/claude-opus-4.6	eng→dyu	0.0	19.1	reported
anthropic/claude-sonnet-4	eng→dyu	2.1	23.7	reported
anthropic/claude-sonnet-4	dyu→eng	1.6	18.8	reported
anthropic/claude-sonnet-4.5	eng→dyu	2.3	23.5	reported
anthropic/claude-sonnet-4.5	dyu→eng	4.3	23.7	reported
google/gemini-2.5-flash	dyu→eng	8.4	30.6	reported
google/gemini-2.5-flash	eng→dyu	1.5	22.1	reported
google/gemini-3-pro-preview	dyu→eng	16.9	41.1	reported
google/gemini-3-pro-preview	eng→dyu	2.4	24.9	reported

GATONES

Model	direction	BLEU	chrF	Source
google/madlad400-3b-mt	dyu→eng	7.6	28	reported
google/madlad400-3b-mt	eng→dyu	1.9	19.4	reported
google/madlad400-7.2b-mt	dyu→eng	6.7	27.7	reported
google/madlad400-7.2b-mt	eng→dyu	2.3	19.6	reported
google/madlad400-10.7b-mt	dyu→eng	8.6	30	reported
google/madlad400-10.7b-mt	eng→dyu	2.2	20.1	reported
google/madlad400-8b-lm	dyu→eng	1.3	15.1	reported
google/madlad400-8b-lm	eng→dyu	0	1.5	reported

FLORES-200

Model	direction	BLEU	chrF	Source
google/madlad400-3b-mt	dyu→eng	7.9	31.5	reported
google/madlad400-3b-mt	eng→dyu	2	21	reported
google/madlad400-7.2b-mt	dyu→eng	7.9	31.3	reported
google/madlad400-7.2b-mt	eng→dyu	2.3	21.1	reported
google/madlad400-10.7b-mt	dyu→eng	8.9	33.5	reported
google/madlad400-10.7b-mt	eng→dyu	2.1	22.1	reported
google/madlad400-8b-lm	dyu→eng	2.3	18.2	reported
google/madlad400-8b-lm	eng→dyu	0	5.8	reported

LLM (Large Language Models)

FLORES-200

Model	Log-perplexity	Source
goldfish-models/dyu_latn_full	183.05	reported

SIB-200 10-sample

Model	Classification Acc.	Source
anthropic/claude-opus-4.5	100.0	reported
google/gemini-2.5-pro	100.0	reported
meta-llama/llama-3.1-70b-instruct	90.0	reported

ASR Models (7)

Model	Size	Downloads	Likes
espnet/xeus	—	60	146
Klayt/xlsr-dioula-asr	0.3B	30	0
johntsi/ZeroSwot-Medium_asr-cv_en-to-200	0.4B	17	0
johntsi/ZeroSwot-Large_asr-cv_en-to-200	0.4B	16	1
UBC-NLP/Simba-SLID-49	94.6M	15	0
johntsi/ZeroSwot-Medium_asr-mustc_en-to-200	0.4B	14	0
johntsi/ZeroSwot-Large_asr-mustc_en-to-200	0.4B	12	0

View on HuggingFace: dyu: 7

TTS Models (23)

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
bilalRHCH/Arabic-text-to-speech	—	14	0
phucpx247/OmniVoice	—	14	0
gubernac/OmniVoice	—	13	1
preference1/OmniVoice	—	13	0
EBINSHAJI1999/OmniVoice	—	12	0
Citrus-777/OmniVoice	—	12	0

View on HuggingFace: dyu: 23

MT Models (30)

Model	Size	Downloads	Likes
facebook/nllb-200-distilled-600M	—	1.1M	880
google/madlad400-3b-mt	3B	329.0K	192
facebook/nllb-200-distilled-1.3B	—	147.0K	181
facebook/nllb-200-3.3B	—	81.6K	443
jbochi/madlad400-3b-mt	3B	76.4K	154
facebook/nllb-200-1.3B	—	31.4K	71
Xenova/nllb-200-distilled-600M	—	4.4K	50
JustFrederik/nllb-200-distilled-600M-ct2-int8	—	925	1
facebook/nllb-moe-54b	—	557	123
JustFrederik/nllb-200-1.3B-ct2-int8	—	200	0
JustFrederik/nllb-200-distilled-600M-ct2-float16	—	157	0
winstxnhdw/nllb-200-distilled-1.3B-ct2-int8	—	141	6
JustFrederik/nllb-200-distilled-1.3B-ct2-int8	—	116	2
JustFrederik/nllb-200-distilled-600M-ct2	—	65	2
JustFrederik/nllb-200-3.3B-ct2-float16	—	52	3
michaelfeil/ct2fast-nllb-200-distilled-1.3B	—	47	1
michaelfeil/ct2fast-nllb-200-3.3B	—	43	14
ychenNLP/nllb-200-3.3B-easyproject	—	36	8
JustFrederik/nllb-200-distilled-1.3B-ct2-float16	—	30	1
JustFrederik/nllb-200-distilled-1.3B-ct2	—	29	0

View on HuggingFace: dyu: 30

LLM Models (5)

Model	Size	Downloads	Likes
google/madlad400-8b-lm	9B	183	11
jbochi/madlad400-8b-lm	9B	14	8
goldfish-models/dyu_latn_10mb	—	8	0
goldfish-models/dyu_latn_5mb	—	8	0
goldfish-models/dyu_latn_full	—	1	0

View on HuggingFace: dyu: 5

Datasets

ASR Datasets (10)

Dataset	Rows	Downloads	Likes
fsicoli/common_voice_22_0	—	21.2K	16
fsicoli/common_voice_15_0	—	8.4K	6
fsicoli/common_voice_16_0	—	5.1K	4
fsicoli/common_voice_17_0	—	3.7K	12
fsicoli/common_voice_21_0	—	3.2K	2
UBC-NLP/SimbaBench_dataset	748.0K	975	0
fsicoli/common_voice_19_0	—	872	7
WueNLP/sib-fleurs	94.4K	260	11
fsicoli/common_voice_18_0	—	210	9
Snagy22000/common_voice_17_0	—	86	0

View on HuggingFace: dyu: 10

TTS Datasets (3)

Dataset	Rows	Downloads	Likes
sarulab-speech/commonvoice22_sidon	15.1M	1.1K	29
UBC-NLP/SimbaBench_dataset	748.0K	975	0
WueNLP/sib-fleurs	94.4K	260	11

View on HuggingFace: dyu: 3

Translation Datasets (17)

Dataset	Rows	Downloads	Likes
facebook/flores	—	43.4K	97
openlanguagedata/flores_plus	893.0K	15.8K	124
google/smol	798.0K	2.8K	94
Muennighoff/flores200	—	1.6K	20
ayymen/Pontoon-Translations	3.6M	1.1K	18
mteb/FloresBitextMining	2.0K	303	1
visheratin/laion-coco-nllb	894.0K	255	44
slone/nllb-200-10M-sample	10.0M	168	13
community-datasets/udhr	—	123	2
mteb/flores	2.0K	96	0
FrancophonIA/gatitos	—	61	4
vpermilp/nllb-200-distilled-600M-rust	—	51	1
robinhad/long_flores	—	49	0
Svngoku/xP3x-Kongo	1.2M	30	4
neulab/PangeaBench-flores	—	17	0
Bretagne/PolyWrite_clean	155	11	0
SupriyoCode/flores200	—	11	0

View on HuggingFace: dyu: 17