Dataset Viewer
Auto-converted to Parquet Duplicate
sentence_id
int32
1.57M
12.6M
text_raw
stringlengths
3
198
text
stringlengths
3
198
audio
audioduration (s)
0.79
12.7
character_fixed
bool
2 classes
7,536,889
Ɣef leḥsab-nkent, yezmer ad iyi-d-iẓer yiwen?
Ɣef leḥsab-nkent, yezmer ad iyi-d-iẓer yiwen?
false
7,276,328
Tzemreḍ ad iyi-tawiḍ ɣer unafag, ma ulac aɣilif?
Tzemreḍ ad iyi-tawiḍ ɣer unafag, ma ulac aɣilif?
false
7,299,697
Tḥebseḍ Tom?
Tḥebseḍ Tom?
false
7,204,695
Yella win i yi-d-inulen.
Yella win i yi-d-inulen.
false
8,637,290
Tom yelli isubar.
Tom yelli isubar.
false
7,428,566
Di leɛnaya-k xdem-it.
Di leɛnaya-k xdem-it.
false
10,634,140
Qimemt deg lqaɛa!
Qimemt deg lqaɛa!
false
9,753,328
Nekni akk seg Boston.
Nekni akk seg Boston.
false
9,855,709
Tuɣeḍ-d ayefki?
Tuɣeḍ-d ayefki?
false
7,314,973
Tḥemmleḍ tamwansa n Tom?
Tḥemmleḍ tamwansa n Tom?
false
9,542,114
Meqqret aṭas fell-awen.
Meqqret aṭas fell-awen.
false
7,449,854
Tfeṛḥem s wayenni?
Tfeṛḥem s wayenni?
false
8,059,415
Ur ttekkiɣ ara deg uqisus-ayi.
Ur ttekkiɣ ara deg uqisus-ayi.
false
7,253,882
Nɣellet.
Nɣellet.
false
7,755,375
Ad xedmen ayen i asen-d-nniɣ.
Ad xedmen ayen i asen-d-nniɣ.
false
7,497,402
Sneɣ anagar tafṛansist akked teglizit.
Sneɣ anagar tafṛansist akked teglizit.
false
9,379,800
Nekni seg Boston.
Nekni seg Boston.
false
7,033,516
Tibḥirt-ines d leqdic n tẓuṛi.
Tibḥirt-ines d leqdic n tẓuṛi.
false
7,880,384
Bɣiɣ ad ruḥeɣ tura ɣer uxxam.
Bɣiɣ ad ruḥeɣ tura ɣer uxxam.
false
11,691,685
Tura s tefransist.
Tura s tefransist.
false
8,465,210
D kečč i yessawlen i temsulta?
D kečč i yessawlen i temsulta?
false
9,767,733
Ssewweɣ-awen-tt id i kenwi kan.
Ssewweɣ-awen-tt id i kenwi kan.
false
7,428,855
Yella kra i tettuḍ?
Yella kra i tettuḍ?
false
7,456,576
Yessen Tom abrid?
Yessen Tom abrid?
false
7,234,342
Ilaq ad tettem tiremt n ssbeḥ mkul ass.
Ilaq ad tettem tiremt n ssbeḥ mkul ass.
false
7,610,291
Tom yezmer yerbeḥ.
Tom yezmer yerbeḥ.
false
7,272,365
Ur zmireɣ ara dduɣ ad εummeɣ yid-wen ass-a.
Ur zmireɣ ara dduɣ ad ɛummeɣ yid-wen ass-a.
true
9,445,978
Gerrzen.
Gerrzen.
false
7,527,718
Ddem ayen tuḥwaǧeḍ.
Ddem ayen tuḥwaǧeḍ.
false
9,457,230
Nugad fell-ak.
Nugad fell-ak.
false
9,798,474
D aḥulfu amyaɣ.
D aḥulfu amyaɣ.
false
7,204,769
Lḥiɣ iman-iw.
Lḥiɣ iman-iw.
false
9,678,026
Imir-n bɣiɣ ad ẓṛeɣ ayen s timmad-iw.
Imir-n bɣiɣ ad ẓṛeɣ ayen s timmad-iw.
false
7,826,438
Dima ttfadeɣ.
Dima ttfadeɣ.
false
8,685,017
Yedduri Tom?
Yedduri Tom?
false
7,489,332
Yeqqim Tom deg uxxam.
Yeqqim Tom deg uxxam.
false
9,080,608
Rsemt da!
Rsemt da!
false
7,268,614
Yella ḥedd i yeldin tawwurt.
Yella ḥedd i yeldin tawwurt.
false
11,070,218
Tettbanem-d tceɣlem mliḥ.
Tettbanem-d tceɣlem mliḥ.
false
7,188,228
Ɛawnem Tom.
Ɛawnem Tom.
false
7,751,651
Ɣas in-as dayen yekfa wass.
Ɣas in-as dayen yekfa wass.
false
9,470,045
Ɛiwdemt xemmemt, ttxil-kent.
Ɛiwdemt xemmemt, ttxil-kent.
false
8,350,990
Cukkeɣ ad naf Tom.
Cukkeɣ ad naf Tom.
false
7,463,156
Am kečč i nneqmaseɣ.
Am kečč i nneqmaseɣ.
false
9,389,154
Ilaq ad yecnu yid-neɣ Tom.
Ilaq ad yecnu yid-neɣ Tom.
false
8,544,761
Yessen nezzeh ad iwet apyanu.
Yessen nezzeh ad iwet apyanu.
false
7,465,019
Ɣur-neɣ tiḥeṛṛit.
Ɣur-neɣ tiḥeṛṛit.
false
8,325,359
Ur yeεlim Tom d acu nexdem.
Ur yeɛlim Tom d acu nexdem.
true
7,453,772
Ɣṛet-it-id i tikkelt nniḍen.
Ɣṛet-it-id i tikkelt nniḍen.
false
7,188,035
Ččet s leεqel-nwen.
Ččet s leɛqel-nwen.
true
9,642,527
La d-akent-deɛɛuɣ.
La d-akent-deɛɛuɣ.
false
7,577,955
Kra yellan d tafat tella tecεel.
Kra yellan d tafat tella tecɛel.
true
8,435,436
Teskaddbeḍ i imawlan-ik?
Teskaddbeḍ i imawlan-ik?
false
9,688,249
Eǧǧ-iyi ad ak-tt-id xedmeɣ.
Eǧǧ-iyi ad ak-tt-id xedmeɣ.
false
10,007,784
Ẓriɣ beddleɣ.
Ẓriɣ beddleɣ.
false
8,065,114
Aql-iyi d tawaziwt.
Aql-iyi d tawaziwt.
false
7,358,415
Xdem sin ijeṛṛiḍen.
Xdem sin ijeṛṛiḍen.
false
7,277,622
Ilaq ad yili kra n ḥedd ara ixedmen kra.
Ilaq ad yili kra n ḥedd ara ixedmen kra.
false
7,304,672
Beṛka-kem amennuɣ.
Beṛka-kem amennuɣ.
false
9,766,590
Nekk d ameεdaz mliḥ.
Nekk d ameɛdaz mliḥ.
true
7,347,160
Ur bɣiɣ ara ad yi-d-teẓreḍ d taεeryant.
Ur bɣiɣ ara ad yi-d-teẓreḍ d taɛeryant.
true
7,273,930
Sriḥemt aya.
Sriḥemt aya.
false
9,069,117
Kkes ugur.
Kkes ugur.
false
7,425,939
Yella ḥedd i yeṭṭalayen Tom.
Yella ḥedd i yeṭṭalayen Tom.
false
7,316,425
Ur ttmeslayemt ara!
Ur ttmeslayemt ara!
false
8,863,768
Ttwakerrceɣ kra n tikwal.
Ttwakerrceɣ kra n tikwal.
false
7,496,387
Bɣiɣ ad rnuɣ ad sleɣ.
Bɣiɣ ad rnuɣ ad sleɣ.
false
9,092,308
Ansayen yelhan yessefk ad ttwaḥerzen.
Ansayen yelhan yessefk ad ttwaḥerzen.
false
8,309,451
Qrib ttuɣ.
Qrib ttuɣ.
false
10,788,933
Ɣef wacu tecfiḍ?
Ɣef wacu tecfiḍ?
false
9,759,216
A win yufan ad tɛiwnemt timdukkal-nkent.
A win yufan ad tɛiwnemt timdukkal-nkent.
false
10,586,936
Ɣef leḥsab-nkent, Tom ad yexdem aya?
Ɣef leḥsab-nkent, Tom ad yexdem aya?
false
9,417,033
Yusa-d yewεer lḥal tazwara.
Yusa-d yewɛer lḥal tazwara.
true
8,903,885
Tura kan i d-uwḍeɣ.
Tura kan i d-uwḍeɣ.
false
7,576,739
Abrid yegres, ihi ɣas ḥader..
Abrid yegres, ihi ɣas ḥader..
false
9,638,737
D acu taḥwaǧeḍ?
D acu taḥwaǧeḍ?
false
11,144,640
Meqqret tcehrit-is.
Meqqret tcehrit-is.
false
10,594,378
Yekṛeh Tom tugna-a.
Yekṛeh Tom tugna-a.
false
7,057,046
Telliḍ tusmeḍ, anaɣ?
Telliḍ tusmeḍ, anaɣ?
false
7,188,427
Beṛka-kem aɣenni.
Beṛka-kem aɣenni.
false
7,457,109
Tebɣiḍ ad ak-t-id ɣṛeɣ?
Tebɣiḍ ad ak-t-id ɣṛeɣ?
false
10,688,114
Mazal ur fukkeɣ.
Mazal ur fukkeɣ.
false
7,971,610
Iḍ-ayi ara bduɣ.
Iḍ-ayi ara bduɣ.
false
11,058,078
Yeffeɣ ddabex ɣer berra.
Yeffeɣ ddabex ɣer berra.
false
7,316,404
Ḥṛes afus-iw.
Ḥṛes afus-iw.
false
7,264,455
Yella kra i teččiḍ ṣṣbeḥ-agi?
Yella kra i teččiḍ ṣṣbeḥ-agi?
false
9,531,602
Zemreɣ ad k-ḍemneɣ.
Zemreɣ ad k-ḍemneɣ.
false
9,521,349
Aql-aɣ ad d-nas i lmendad-ik.
Aql-aɣ ad d-nas i lmendad-ik.
false
7,465,046
Ala, tenemmirt. Ṛwiɣ.
Ala, tenemmirt. Ṛwiɣ.
false
10,581,003
Amek armi i temyussanem?
Amek armi i temyussanem?
false
10,639,059
Iḥemmel Tom aguglu n Leswis.
Iḥemmel Tom aguglu n Leswis.
false
8,989,778
Tudert d nnuba n ṭṭrenǧ.
Tudert d nnuba n ṭṭrenǧ.
false
7,087,422
D wanwa tmeslayeḍ ?
D wanwa tmeslayeḍ ?
false
7,188,450
Tixeṛ-ik ur ttɛeyyiḍ ara.
Tixeṛ-ik ur ttɛeyyiḍ ara.
false
8,963,943
Teccfeḍ assa?
Teccfeḍ assa?
false
11,180,388
Myecmummaḥen.
Myecmummaḥen.
false
9,704,959
Ḍebbeṛ aqeṛṛu-ik.
Ḍebbeṛ aqeṛṛu-ik.
false
9,498,183
D acu i as-d-tuɣemt i Tom i Newwal?
D acu i as-d-tuɣemt i Tom i Newwal?
false
7,058,160
Ifaz!
Ifaz!
false
10,677,226
Bɣiɣ ad yemmag waya.
Bɣiɣ ad yemmag waya.
false
End of preview. Expand in Data Studio

Tatoeba Kabyle Audio Dataset

A clean, standardized audio-text dataset for Kabyle (Taqbaylit) automatic speech recognition, extracted from the Tatoeba Project and rigorously orthographically corrected.

Dataset Description

This dataset contains 47,789 Kabyle sentences with audio recordings (~25.78 hours total) sourced from Tatoeba. All transcriptions have been standardized to use correct Kabyle Latin characters, replacing visually similar false friends from Greek, Cyrillic, and other Latin scripts that commonly contaminate Kabyle text corpora.

Character Standardization

Kabyle uses the following Latin characters in its standard orthography:

  • ɛ (U+025B) / Ɛ (U+0190) — open-mid front unrounded vowel
  • ɣ (U+0263) / Ɣ (U+0194) — voiced velar fricative
  • č (U+010D) / Č (U+010C) — voiceless postalveolar affricate
  • ǧ (U+01E7) / Ǧ (U+01E6) — voiced postalveolar affricate
  • (U+1E25), (U+1E0D), (U+1E5B), (U+1E6D), (U+1E63), (U+1E93) — emphatic/pharyngeal consonants

These are frequently confused with visually similar characters in raw web data. This dataset applies the following corrections:

False Friend Correct Kabyle Unicode Origin
ε (Greek epsilon) ɛ U+025B Greek
Σ (Greek sigma) Ɛ U+0190 Greek
γ (Greek gamma) ɣ U+0263 Greek
Γ (Greek Gamma) Ɣ U+0194 Greek
Ԑ (Cyrillic rev. Ze) Ɛ U+0190 Cyrillic
ԑ (Cyrillic rev. ze) ɛ U+025B Cyrillic
З (Cyrillic Ze) Ɛ U+0190 Cyrillic
з (Cyrillic ze) ɛ U+025B Cyrillic
Ǝ (Latin reversed E) Ɛ U+0190 Latin
ǝ (Latin turned e) ɛ U+025B Latin
ż / Ż (Polish) ẓ / Ẓ U+1E93 / U+1E92 Latin
ṙ / Ṙ (Irish) ṛ / Ṛ U+1E5B / U+1E5A Latin
ṫ / Ṫ ṭ / Ṭ U+1E6D / U+1E6C Latin
ṡ / Ṡ (Irish) ṣ / Ṣ U+1E63 / U+1E62 Latin
ḋ / Ḋ (Irish) ḍ / Ḍ U+1E0D / U+1E0C Latin
ḣ / Ḣ ḥ / Ḥ U+1E25 / U+1E24 Latin
ċ / Ċ (Maltese) č / Č U+010D / U+010C Latin
ć / Ć (Polish/Croatian) č / Č U+010D / U+010C Latin
ç / Ç (French) č / Č U+010D / U+010C Latin
ĉ / Ĉ (Esperanto) č / Č U+010D / U+010C Latin
ġ / Ġ (Maltese) ǧ / Ǧ U+01E7 / U+01E6 Latin
ǥ / Ǥ (Skolt Sami) ǧ / Ǧ U+01E7 / U+01E6 Latin
ğ / Ğ (Turkish) ǧ / Ǧ U+01E7 / U+01E6 Latin
ĝ / Ĝ (Esperanto) ǧ / Ǧ U+01E7 / U+01E6 Latin

Additionally, Unicode NFC normalization is applied to resolve precomposed vs. decomposed forms (e.g., z + combining dot below → ).

3,643 sentences (7.62%) required at least one character fix.

Audio Statistics

Metric Value
Total clips 47,789
Total duration 25.78 hours
Sampling rate 16 kHz mono
Mean duration 1.94 seconds
Median duration 1.82 seconds
Min duration 0.73 seconds
Max duration 12.65 seconds

The audio consists of short, read-aloud sentences typical of the Tatoeba project. The tight duration distribution (most clips are 1–3 seconds) makes this dataset well-suited for Wav2Vec 2.0 CTC training with minimal memory overhead.

File Sizes

Split Size
train ~807 MB
validation ~43 MB
Total download ~803 MB
Total dataset ~851 MB

Splits

Split Examples
train 45,399
validation 2,390

Features

  • sentence_id: Tatoeba sentence ID
  • text_raw: Original transcription from Tatoeba (before cleaning)
  • text: Cleaned, standardized Kabyle transcription
  • audio: Audio waveform (16 kHz, mono)
  • character_fixed: Boolean flag indicating whether a false-friend fix was applied

Usage

from datasets import load_dataset

ds = load_dataset("boffire/tatoeba-kabyle-audio")
sample = ds["train"][0]
print(sample["text"])
# → standardized Kabyle sentence

Citation

If you use this dataset, please cite the Tatoeba Project:

@misc{tatoeba,
  title = {Tatoeba: Collection of sentences and translations},
  howpublished = {\url{https://tatoeba.org}},
}

License

Audio files and sentences from Tatoeba are licensed under CC BY 4.0 (or compatible individual licenses as noted in the original Tatoeba metadata).

Acknowledgments

Downloads last month
73