Datasets:

CLEAR-Global
/

BibleTTS-Hausa

text stringlengths 3 421	duration float32 0.32 43.5	score float64 -54.97 -0.02	speaker stringclasses 2 values
rubutaccen tarihi daga adamu zuwa ibrahim.	4.02	-7.209332	two
zuwa ’ya’yan nuhu maza.	1.96	-32.801132	two
adamu, set, enosh, kenan, mahalalel, yared, enok, metusela, lamek, nuhu.	11.54	-13.782688	two
’ya’yan nuhu maza su ne, shem, ham da yafet.	5.02	-27.571634	two
mutanen yafet.	1.64	-13.64579	two
’ya’yan yafet maza su ne, gomer, magog, madai, yaban, tubal, meshek da tiras.	9.38	-26.539461	two
’ya’yan gomer maza su ne, ashkenaz, rifat da togarma.	6.52	-30.159201	two
’ya’yan yaban maza su ne, elisha, tarshish, kittim da rodanim.	7.9	-39.82951	two
mutanen ham.	1.68	-4.702216	two
’ya’yan ham maza su ne, kush, masar fut da kan’ana.	6.32	-26.596724	two
’ya’yan kush maza su ne, seba, hawila, sabta, ra’ama da sabteka.	8.26	-14.731167	two
’ya’yan ra’ama maza su ne, sheba da dedan.	4.92	-7.286259	two
kush shi ne mahaifin nimrod, wanda ya yi girma ya zama babban jarumi a duniya.	7.9	-22.940752	two
masar shi ne mahaifin ludiyawa, anamawa, lehabiyawa, neftuhawa, fatrusawa, kasluhiyawa daga waɗanda filistiyawa suka fito da kaftorawa.	16.16	-34.652424	two
kan’ana shi ne mahaifin sidon ɗan farinsa, da hittiyawa, yebusiyawa, amoriyawa, girgashiyawa.	10.8	-25.103592	two
hiwiyawa, arkiyawa, siniyawa, arbadiyawa, zemarawa da hamawa.	9.34	-7.022654	two
semiyawa.	1.64	-1.253726	two
’ya’yan shem maza su ne, elam, asshur, arfakshad, lud da aram.	7.9	-44.777458	two
’ya’yan aram maza su ne, uz, hul, geter da meshek.	5.42	-20.578074	two
arfakshad shi ne mahaifin shela, shela kuwa shi ne mahaifin eber.	5.8	-13.195524	two
aka haifa wa eber ’ya’ya maza biyu.	3.16	-9.279648	two
aka kira ɗaya feleg domin a lokacinsa ne aka raba duniya; sunan ɗan’uwansa kuwa shi ne yoktan.	8.18	-28.07671	two
yoktan shi ne mahaifin almodad, shelef, hazarmawet, yera.	7.32	-21.83951	two
hadoram, uzal, dikla, ebal, abimayel, sheba, ofir, hawila da yobab.	11.16	-16.838064	two
dukan waɗannan ’ya’yan yoktan maza ne.	3.9	-18.999725	two
zuriyar shem.	1.86	-7.986124	two
shem, arfakshad, shela, eber, feleg, reyu.	7.04	-21.994097	two
serug, nahor, tera.	3.52	-3.275241	two
da abram wato, ibrahim.	3	-2.737046	two
iyalin ibrahim.	1.84	-5.617513	two
’ya’yan ibrahim maza su ne, ishaku da ishmayel.	4.5	-28.108051	two
zuriyar hagar.	1.34	-1.462367	two
waɗannan su ne zuriyarsu.	1.74	-11.082809	two
nebayiwot ɗan farin ishmayel, kedar, adbeyel, mibsam, mishma, duma, massa, hadad, tema.	11.78	-26.219616	two
yetur, nafish da kedema.	3.6	-3.725911	two
waɗannan su ne ’ya’yan ishmayel maza.	3.04	-15.321421	two
zuriyar ketura.	1.66	-5.1129	two
’ya’yan ketura maza, ƙwarƙwarar ibrahim su ne, zimran, yokshan, medan, midiyan, ishbak da shuwa.	9.74	-20.392855	two
’ya’yan yokshan maza su ne, sheba da dedan.	4.98	-11.039001	two
’ya’yan midiyan maza su ne, efa, efer, hanok, abida da elda’a.	7.94	-23.202824	two
dukan waɗannan su ne zuriyar ketura.	2.8	-7.625967	two
zuriyar saratu.	1.64	-1.825044	two
ibrahim shi ne mahaifin ishaku.	2.3	-2.288441	two
’ya’yan ishaku maza su ne, isuwa da isra’ila.	3.8	-10.738405	two
’ya’yan isuwa maza.	2.8	-13.44835	two
’ya’yan isuwa maza, su ne, elifaz, reyuwel, yewush, yalam da kora.	8.04	-23.243996	two
’ya’yan elifaz maza su ne, teman, omar, zefi, gatam, kenaz, da timna wanda aka haifa wa amalek.	10.78	-25.458977	two
’ya’yan maza reyuwel su ne, nahat, zera, shamma da mizza.	7.26	-15.157611	two
mutanen seyir a edom.	2.98	-4.322958	two
’ya’yan seyir maza su ne, lotan, shobal, zibeyon, ana, dishon, ezer da dishan.	9.22	-27.458328	two
’ya’yan lotan maza su ne, hori da homam.	4.22	-9.441706	two
timna ita ce ’yar’uwar lotan.	2.96	-7.367174	two
’ya’yan shobal maza su ne, alwan, manahat, ebal, shefo da onam.	7.46	-12.225489	two
’ya’yan zibeyon maza su ne, aiya da ana.	4.4	-24.145027	two
ɗan ana shi ne, dishon.	2.14	-2.674628	two
’ya’yan dishon maza su ne, hemdan, eshban, itran da keran.	6.2	-29.70134	two
’ya’yan ezer maza su ne, bilhan, za’aban da ya’akan.	6.42	-34.452816	two
’ya’yan dishan maza su ne, uz da aran.	4.42	-22.065636	two
masu mulkin edom.	2.04	-3.835129	two
waɗannan su ne sarakunan da suka yi sarauta a edom kafin wani sarkin mutumin isra’ila yă yi sarauta.	6.64	-18.074001	two
bela ɗan beyor, wanda aka kira birninsa dinhaba.	4.64	-20.361496	two
sa’ad da bela ya mutu sai yobab ɗan zera daga bozra ya gāje shi a matsayin sarki.	8.42	-9.285899	two
da yobab ya mutu, husham daga ƙasar temaniyawa ya gāje shi a matsayin sarki.	7.36	-12.662523	two
sa’ad da husham ya mutu, hadad ɗan bedad, wanda ya ci midiyan da yaƙi a cikin ƙasar mowab, ya gāje shi a matsayin sarki.	9.9	-16.145393	two
aka kira birninsa awit.	2.52	-2.529839	two
da hadad ya mutu, samla daga masreka ya gāje shi a matsayin sarki.	6.48	-5.821884	two
sa’ad da samla ya mutu, sha’ul daga rehobot na kogi ya gāje shi a matsayin sarki.	7.64	-15.848474	two
sa’ad da sha’ul ya mutu, ba’al hanan ɗan akbor ya gāje shi a matsayin sarki.	7.32	-10.207342	two
sa’ad da ba’al hanan ya mutu, hadad ya gāje shi a matsayin sarki.	5.2	-7.94296	two
aka kira birninsa fau, sunan matarsa kuwa mehetabel ’yar matired, ’yar me zahab.	9.04	-26.939987	two
hadad shi ma ya mutu.	2.34	-2.225162	two
manyan edom su ne, timna, alwa, yetet.	4.86	-9.641237	two
oholibama, ela, finon.	3.78	-6.143802	two
kenaz, teman, mibzar, magdiyel da iram.	6.14	-7.182587	two
waɗannan su ne manyan edom.	2.28	-13.678022	two
daya tarihi sura biyu.	2.8	-46.647022	two
’ya’yan isra’ila maza.	2.28	-11.068502	two
waɗannan su ne ’ya’yan isra’ila maza.	2.86	-12.605809	two
ruben, simeyon, lawi, yahuda, issakar, zebulun, dan, yusuf, benyamin, naftali, gad da asher.	11.44	-29.856564	two
zuriyar yahuda.	1.9	-13.035345	two
’ya’yan hezron maza.	2.24	-32.090664	two
’ya’yan yahuda maza su ne, er, onan da shela.	4.24	-23.127348	two
waɗannan mutum uku bat shuwa mutuniyar kan’ana ce ta haifa masa.	5.32	-12.753726	two
er, ɗan farin yahuda mugu ne a gaban ubangiji, saboda haka ubangiji ya kashe shi.	5.74	-19.332855	two
tamar, surukar yahuda, ta haifa masa ferez da zera.	4.82	-5.956216	two
yahuda ya haifi ’ya’ya maza biyar ne.	3	-26.35009	two
’ya’yan ferez maza su ne, hezron da hamul.	4.58	-6.5176	two
’ya’yan zera maza su ne, zimri, etan, heman, kalkol da darda, su biyar ne.	7.32	-13.988387	two
’ya’yan karmi maza su ne, akar wanda ya kawo masifa wa isra’ila ta wurin yin abin da aka haramta.	8.88	-7.803909	two
ɗan etan shi ne, azariya.	2.5	-1.970571	two
’ya’ya maza da aka haifa wa hezron su ne, yerameyel, ram da kaleb.	6.64	-19.725475	two
daga ram ɗan hezron.	2.78	-1.098549	two
ram shi ne mahaifin amminadab, amminadab kuwa shi ne mahaifin nashon, shugaban mutanen yahuda.	8.16	-10.136009	two
nashon shi ne mahaifin salma, salma shi ne mahaifin bowaz, bowaz shi ne mahaifin obed kuma obed shi ne mahaifin yesse.	10.52	-29.358595	two
yesse shi ne mahaifin.	1.62	-13.485482	two
eliyab ɗan farinsa, ɗansa na biyu shi ne abinadab, na ukun shimeya, na huɗun netanel, na biyar raddai, na shidan ozem na bakwai kuma dawuda.	15.94	-16.275536	two
’yan’uwansu mata su ne zeruhiya da abigiyel.	4.44	-13.543561	two
’ya’yan zeruhiya maza guda uku su ne abishai, yowab da asahel.	5.44	-15.997648	two
abigiyel ita ce mahaifiyar amasa, wanda mahaifinsa shi ne yeter mutumin ishmayel.	6.24	-17.132717	two
zuriyar kaleb ɗan hezron.	2.22	-12.521272	two

End of preview. Expand in Data Studio

BibleTTS Hausa — Aligned Bible Speech

101.5 hours of studio-quality Hausa Bible narration from two speakers, force-aligned and segmented at verse level into 43,405 clips (24 kHz, mono, WAV).

Languages

hau — Hausa

Dataset Structure

Audio is stored in the audio column (HuggingFace Audio feature). Parquet shards live under data/read/, one or more files per split.

from datasets import load_dataset

ds = load_dataset("CLEAR-Global/BibleTTS-Hausa", "read", split="train")

field	description
`audio`	The clip. `audio.path` is the source file name, `<BOOK>_<chapter>_<segment>.wav`.
`text`	Verse transcript, lowercase.
`duration`	Clip duration in seconds.
`score`	Alignment confidence for the clip (see below).
`speaker`	Speaker label, `one` or `two`. The corpus is read by two narrators, each covering a distinct set of books.

Statistics

split	clips	hours	median / p95 duration (s)
`train`	40,832	92.68	7.4 / 17.0
`dev`	301	0.78	8.6 / 20.7
`test`	127	0.24	6.2 / 14.4
`other`	2,145	7.76	9.6 / 37.1
total	43,405	101.46	7.4 / 18.0

The corpus has two speakers, and they are split by book rather than interleaved: speaker one reads 28 books (22,487 clips, 56.40 h) and speaker two reads 38 books (20,918 clips, 45.06 h). No clip mixes speakers. Beyond the two labels there is no speaker or demographic metadata.

Splits and the Alignment Score

Each clip carries a score from the forced aligner: the alignment confidence for that clip, expressed as an average per-token log-probability. Values are negative; higher is better. It is length-normalised, so long and short clips are comparable.

The corpus ships with three authoritative splits. They are held out at book level: dev draws entirely from the book of Ezra and test entirely from the book of Colossians, so evaluation is on unseen books rather than on held-out utterances from a book the model has already seen.

train / dev / test — 41,260 clips (93.70 h), the split assignment shipped with this release.
other — 2,145 clips (7.76 h) that appear in the corpus's full manifest but in none of the three splits. Every one has an alignment score at or below -55 (median -73.6, minimum -721.1), so they are published here for completeness and are not intended for TTS training as-is.

Dataset Creation

The Hausa subset of BibleTTS is derived from the Bible recordings released by Open.Bible and published by Biblica. CLEAR Global produced the first BibleTTS Hausa release by force-aligning those recordings to the Hausa Bible text, but its segmentation contained many alignment errors. This release force-aligns and re-segments the same recordings again, yielding the per-clip transcripts, durations and alignment scores used here. Clips are 24 kHz mono PCM WAV.

Related resources

This corpus was used to train the Hausa text-to-speech models in the Hausa TTS models collection (VITS, XTTS, YourTTS).

Those TTS models were then used to synthesise additional Hausa speech, and that synthetic audio was used to train some of the models in the Hausa W2v-BERT 2.0 and Hausa MMS-1B collections. This corpus was therefore never used to train ASR models directly, but those models descend from it: the chain is this corpus → TTS models → synthetic speech → ASR models.

Similar datasets from CLEAR Global and collaborators

Hausa Synthetic ASR Dataset (XTTS) — synthetic Hausa speech for ASR generated with XTTS.
Hausa Synthetic ASR Dataset (YourTTS) — synthetic Hausa speech for ASR generated with YourTTS.
TWB Voice Hausa 1.0 — read speech for TTS.

Caveats

Two speakers, Bible-domain vocabulary: not representative of everyday spoken Hausa.
Verse-level segmentation follows the Bible text; some clips are verse fragments, so transcripts are not always complete sentences.
The other split is deliberately unfiltered for alignment quality and contains misaligned clips.

Citation

@inproceedings{meyer22c_interspeech,
  title     = {{BibleTTS: a large, high-fidelity, multilingual, and uniquely African speech corpus}},
  author    = {Josh Meyer and David Adelani and Edresson Casanova and Alp Öktem and Daniel Whitenack and Julian Weber and Salomon {KABONGO KABENAMUALU} and Elizabeth Salesky and Iroro Orife and Colin Leong and Perez Ogayo and Chris {Chinenye Emezue} and Jonathan Mukiibi and Salomey Osei and Apelete AGBOLO and Victor Akinode and Bernard Opoku and Olanrewaju Samuel and Jesujoba Alabi and Shamsuddeen Hassan Muhammad},
  year      = {2022},
  booktitle = {{Interspeech 2022}},
  pages     = {2383--2387},
  doi       = {10.21437/Interspeech.2022-10850},
  issn      = {2958-1796},
}

@inproceedings{derenzi-etal-2025-synthetic,
  title     = "Synthetic Voice Data for Automatic Speech Recognition in {A}frican Languages",
  author    = "DeRenzi, Brian and Dixon, Anna and Farhi, Mohamed Aymane and Resch, Christian",
  booktitle = "Proceedings of the First Workshop on Advancing {NLP} for Low-Resource Languages",
  month     = sep,
  year      = "2025",
  address   = "Varna, Bulgaria",
  publisher = "INCOMA Ltd., Shoumen, Bulgaria",
  url       = "https://aclanthology.org/2025.lowresnlp-1.16/",
  pages     = "152--186"
}

Contact

CLEAR Global — tech@clearglobal.org

Acknowledgments

Force-alignment, segmentation and packaging by CLEAR Global, continuing the Masakhane BibleTTS collaboration. Original recordings from the Open.Bible project by Biblica.

License

CC BY-SA 4.0 (SPDX: cc-by-sa-4.0)

Downloads last month: 42