aoiandroid/mms-multilingual-audio-5to30min

Multilingual Audio Dataset (5-30min)

This dataset contains continuous speech audio files for various languages (ranging from 5 to 30 minutes in length per language) collected from diverse sources including Hugging Face and YouTube.

Dataset Statistics

• Total Languages: 100
• Sources: HF Omnilingual ASR Corpus, YouTube

Language Details

Language Code	Language Name	Source	Duration (seconds)
jpn	Japanese	youtube	1459.84
eng	English	youtube	669.16
spa	Spanish	youtube	490.01
cmn	Chinese, Mandarin	youtube	438.08
hin	Hindi	youtube	841.39
fra	French	youtube	1249.98
swh	Swahili	youtube	605.02
aae	Aae	youtube	314.31
aal	Afade	youtube	946.28
aao	Aao	youtube	324.39
abn	Abua	youtube	301.38
abr	Abron	youtube	396.48
abs	Malay, Ambonese	youtube	439.42
afo	Ajiri	youtube	481.09
ahl	Igo	youtube	444.33
ahs	Ashe	youtube	718.59
ala	Alago	youtube	763.53
amu	Amuzgo, Guerrero	youtube	401.29
anc	Ngas	youtube	310.59
ank	Goemai	youtube	461.39
anp	Angika	youtube	343.93
anw	Anaang	youtube	437.93
aom	Ömie	youtube	527.21
bbu	Kulung	youtube	1488.77
bcs	Hohumono	youtube	588.63
bcy	Bacama	youtube	342.10
bda	Bayot	youtube	398.85
bde	Bade	youtube	761.92
bdm	Buduma	youtube	379.43
bew	Betawi	youtube	669.57
bhb	Bhili	youtube	758.36
bhh	Bukharic	youtube	633.46
bho	Bhojpuri	youtube	1296.59
bhp	Bima	youtube	456.72
bjj	Kanauji	youtube	480.29
bjk	Barok	youtube	421.84
bjn	Banjar	youtube	457.61
bjt	Balanta-Ganja	youtube	472.50
bky	Bokyi	youtube	990.44
bmq	Bomu	youtube	1551.84
bns	Bundeli	youtube	1680.10
bou	Bondei	youtube	785.30
bqg	Bago-Kusuntu	youtube	1671.96
bra	Braj Bhasha	youtube	400.58
brh	Brahui	youtube	378.83
brx	Boro	youtube	710.09
btm	Batak Mandailing	youtube	430.36
bug	Bugis	youtube	501.42
buo	Terei	youtube	501.97
bux	Boghom	youtube	422.77
bwr	Bura-Pabir	youtube	582.33
byc	Ubaghara	youtube	465.71
bys	Burak	youtube	494.94
byx	Qaqet	youtube	560.02
bzw	Basa	youtube	333.33
ccg	Samba Daka	youtube	905.93
cfa	Dikaka	youtube	311.70
cgg	Chiga	youtube	611.82
chq	Chinantec, Quiotepec	youtube	547.65
ckl	Kibaku	youtube	597.14
cky	Cakfem-Mushere	youtube	430.50
cte	Chinantec, Tepinapa	youtube	390.93
ctl	Chinantec, Tlacoatzintepec	youtube	438.51
dbd	Dadiya	youtube	1124.87
dcc	Deccan	youtube	597.31
deg	Degema	youtube	359.89
dgh	Dghwede	youtube	308.98
dje	Zarma	youtube	381.84
dzg	Dazaga	youtube	539.70
ebu	Kiembu	youtube	334.99
ego	Eggon	youtube	358.77
ekr	Yace	youtube	300.98
elm	Eleme	youtube	1415.49
ets	Etsako	youtube	495.94
etu	Ejagham	youtube	506.09
eyo	Keiyo	youtube	715.01
gbm	Garhwali	youtube	1697.74
gbr	Gbagyi	youtube	347.27
gby	Gbari	youtube	319.81
gdf	Guduf-Gava	youtube	308.98
gjk	Koli, Kachi	youtube	418.12
glw	Glavda	youtube	309.01
gol	Gola	youtube	673.61
gom	Konkani, Goan	youtube	1705.02
gur	Farefare	youtube	318.12
guz	Ekegusii	youtube	502.53
gyz	Gyaazi	youtube	803.97
hah	Hahon	youtube	739.34
haw	Hawaiian	youtube	427.87
hbb	Nya Huba	youtube	389.28
her	Herero	youtube	708.76
hia	Lamang	youtube	839.79
hkk	Hunjara-Kaina Ke	youtube	359.72
hla	Halia	youtube	429.44
hno	Hindko, Northern	youtube	467.32
hoj	Haroti	youtube	1181.05
hue	Huave, San Francisco del Mar	youtube	318.09
hul	Vula’a	youtube	504.91
hwo	Hwana	youtube	1059.71
ida	Luidakho-Luisukha-Lutirichi	youtube	340.66

TranslateBlue migrated fixtures (2026-06-19)

Short speech clips and long-form smoke audio migrated from the local TranslateBlue workspace (Data/audio/, Data/datasets/voice-long-form/audio/).

Path on Hub	Count	Description
translateblue-test-list/audio/*.wav	110	Test-list benchmark clips (~5–25 s); see translateblue-test-list/metadata.csv and manifest.jsonl
translateblue-voice-long-form/audio/*.wav	4	~5 min smoke clips (Filipino, Indonesian, Thai, zh-TW)

Language codes for test-list rows follow translateblue-test-list/language_dataset_mapping.json (ISO 639-3 in column iso639_3).

Note: Existing data/{iso639_3}/audio.wav entries remain the original 5–30 minute MMS/YouTube long-form sources and were not overwritten.