Datasets | Mozilla Data Collective

A collection of read speech recordings in Quechua Jauja Wanka.

Common Voice Scripted Speech 26.0 - Quechua Jauja Wanka

License: CC0-1.0

Locale: qxw

Task: ASR

Format: MP3

Size: 262.44 MB

RFE/RL

RFE/RL Kazakh News Text Corpus

Longitudinal Kazakh news corpus from Radio Azattyq (2003-2026) with over 139,000 articles and 35M tokens.

License: CC-BY-NC-SA-4.0

Locale: kk

Task: NLP

Format: TXT

Size: 126.81 MB

MirasAI

Kannada Time Aligned Speech Corpus

A 5-hour Kannada speech dataset with time-aligned transcriptions, designed for ASR, forced alignment, and speech research.

License: CC-BY-NC-SA-4.0

Locale: kan

Task: ASR

Format: OGG, SRT

Size: 355.77 MB

A collection of read speech recordings in Kihemba.

Common Voice Scripted Speech 26.0 - Kihemba

License: CC0-1.0

Locale: hem

Task: ASR

Format: MP3

Size: 199.58 MB

RFE/RL

RFE/RL Tajik News Text Corpus

Longitudinal Tajik & Russian news corpus from Radio Ozodi (2000-2026) with over 166,000 articles and 20M tokens.

License: CC-BY-NC-SA-4.0

Locale: tg,ru

Task: NLP

Format: TXT

Size: 145.27 MB

Sociocultural dataset for the Gbaya community of Cameroon: identity, kinship, religion, economy, onomastics and a linguistic component.

A Gbaya Sociocultural Dataset

License: NOODL-1.0

Locale: gya

Task: OTH

Format: TSV, PDF

Size: 83.68 KB

A collection of read speech recordings in Erzya.

Common Voice Scripted Speech 26.0 - Erzya

License: CC0-1.0

Locale: myv

Task: ASR

Format: MP3

Size: 65.17 MB

Kokoro Speech Dataset is a public domain Japanese speech dataset. (https://github.com/kaiidams/Kokoro-Speech-Dataset)

Kokoro Speech Dataset

License: libribox

Locale: ja

Task: TTS

Format: FLAC

Size: 3.98 GB

A collection of read speech recordings in Amharic.

Common Voice Scripted Speech 26.0 - Amharic

License: CC0-1.0

Locale: am

Task: ASR

Format: MP3

Size: 59.14 MB

A collection of spontaneous responses to questions in Sabah Malay.

Common Voice Spontaneous Speech 4.0 - Sabah Malay

License: CC0-1.0

Locale: msi

Task: ASR

Format: MP3

Size: 277.94 MB

A collection of read speech recordings in Hausa.

Common Voice Scripted Speech 26.0 - Hausa

License: CC0-1.0

Locale: ha

Task: ASR

Format: MP3

Size: 254.53 MB

This dataset comprises a parallel corpus of 2,154 lines of translations of literary texts from Mada (mxu) to French.

Mada-French Parallel Corpus 1.0

License: NOODL-1.0

Locale: mxu

Task: MT

Format: TSV

Size: 122.37 KB

The dataset contains paired audio and text resources for Teke-Laali, a Bantu language spoken in the Congo. It consists of seven folders containing a total of 9,069 audio clips from raw audio recordings, with a total duration of 7:01:50.126 (HH:MM:SS.mmm). Additionally, there are seven audio/text mapping files containing a total of 9,069 lines. The dataset is suitable for TTS tasks.

Teke-Laali-TTS-Dataset

License: NOODL-1.0

Locale: lli

Task: TTS

Format: WAV, TSV

Size: 635.61 MB

A collection of spontaneous responses to questions in Serian Bidayuh.

Common Voice Spontaneous Speech 4.0 - Serian Bidayuh

License: CC0-1.0

Locale: sdo

Task: ASR

Format: MP3

Size: 201.27 MB

A sentence-level annotated dataset for training and evaluating sequence labeling models such as CRF, BiLSTM, and transformer-based architectures.

Sindhi Word Segmentation (SdSEG)

License: CC-BY-NC-4.0

Locale: snd

Task: NLP

Format: JSON

Size: 1.13 MB

A collection of read speech recordings in IsiNdebele (South).

Common Voice Scripted Speech 26.0 - IsiNdebele (South)

License: CC0-1.0

Locale: nr

Task: ASR

Format: MP3

Size: 869.57 KB

RFE/RL

RFE/RL Turkmen News Text Corpus

Longitudinal Turkmen & Russian news corpus from Azatlyk Radiosy (2009-2026) with nearly 65,000 articles and 16.5M tokens.

License: CC-BY-NC-SA-4.0

Locale: tk,ru

Task: NLP

Format: TXT

Size: 48.28 MB

Segmented Efik speech clips paired with text for TTS. Standard Calabar dialect, Cross River State, Nigeria. 1,362 clips, 5h 32m.

Efik-TTS-Dataset

License: NOODL-1.0

Locale: efi

Task: TTS

Format: MP3, TSV

Size: 297.28 MB

A collection of read speech recordings in Siswati.

Common Voice Scripted Speech 26.0 - Siswati

License: CC0-1.0

Locale: ss

Task: ASR

Format: MP3

Size: 5.85 MB

This dataset contains various topics about everyday life. The topics include society, environment, media, education, culture, health, etc.

Javanese TTS of Banyumasan Dialect

License: CC-BY-SA-4.0

Locale: jav

Task: TTS

Format: WEBM, TSV

Size: 559.08 MB

This dataset comprises audio recordings of scripted speech in Javanese of Lumajang Dialect from East Java of Indonesia.

TTS Javanese-Lumajang Dialect

License: CC-BY-SA-4.0

Locale: jav

Task: TTS

Format: WEBM, TSV

Size: 684.32 MB

CLEAR Global

TWB Parallel Sentence kits - Rohingya (5k)

5,000 English–Rohingya parallel sentences from CLEAR Global's Gamayun mini-kit, translated by professional translators from Tatoeba source sentences.

License: CC-BY-4.0

Locale: rhg

Task: MT

Format: TSV

Size: 358.88 KB