Loading datasets...
Task: ASR
Release Date: 3/22/2026
Format: MP3
Size: 769.10 MB
Share
A collection of read speech recordings in Breton (Brezhoneg).
Restrictions/Special Constraints
None provided.
Forbidden Usage
It is forbidden to attempt to determine the identity of speakers in the Common Voice datasets. It is forbidden to re-host or re-share this dataset.
Intended Use
This dataset is intended to be used for training and evaluating automatic speech recognition (ASR) models. It may also be used for applications relating to computer-aided language learning (CALL) and language or heritage revitalisation.
br)This datasheet is for cv-corpus-25.0-2026-03-09 of the Mozilla Common Voice Scripted Speech dataset for Breton [Brezhoneg - br]. The dataset contains 36377 clips representing 32.89 hours of recorded speech (30.69 hours validated) from 283 speakers, recorded from a text corpus of 41,770 sentences.
Breton (Brezhoneg) is the only Celtic language spoken on the continent. It belongs to the Brittonic branch of the Celtic languages, like Welsh and Cornish. Breton is classified as "severely endangered" by the UNESCO Atlas of the World's Languages in Danger. The language declined from more than one million speakers around 1950 to less than 200 000 in 2024. After a period of intense protest that began in the 1960s, Breton gradually regained a place in schools, on road signs, then in the media, public life in general, and more recently in digital technology. With the creation of the first bilingual schools at the end of the 1970s, young people were once again speaking Breton, and a change in the way people viewed the language and its use began. Approximately 20 000 pupils attend Breton bilingual classes, either in the Diwan network (immersive teaching) or in bilingual classes in public and catholic school systems. In addition to evening breton classes for adults (~4 000), long-term courses (6 to 9 months) are offered to adults, with approximately 350 places available annually and online courses too (more than 15 000 learners).
| Code | Accent | Clips | Speakers |
|---|---|---|---|
| kerneveg | Kerneveg | 9,438 (25.9%) | 29 (10.2%) |
| gwenedeg | Gwenedeg | 2,651 (7.3%) | 14 (4.9%) |
| leoneg | Leoneg | 2,118 (5.8%) | 31 (11.0%) |
| tregerieg | Tregerieg | 255 (0.7%) | 5 (1.8%) |
| - | Other | 99 (0.3%) | 5 (1.8%) |
The dataset includes the following self-declared age and gender distributions. A coverage summary is shown below each table.
Self-declared gender information. The table shows clip and speaker counts with percentages. Speakers who did not declare a gender are listed as Unspecified. A dash (-) indicates zero.
| Code | Gender | Clips | Speakers |
|---|---|---|---|
| male_masculine | Male, masculine | 21,202 (58.3%) | 65 (23.0%) |
| female_feminine | Female, feminine | 2,275 (6.3%) | 17 (6.0%) |
| transgender | Transgender | - | - |
| non-binary | Non-binary | - | - |
| do_not_wish_to_say | Prefer not to say | - | - |
| - | Unspecified | 12,900 (35.5%) | 216 (76.3%) |
Gender declared: 23,477 of 36,377 clips (64.5%), 67 of 283 speakers (23.7%)
Self-declared age information. The table shows clip and speaker counts with percentages. Speakers who did not declare an age are listed as Unspecified. A dash (-) indicates zero.
| Code | Age | Clips | Speakers |
|---|---|---|---|
| teens | Teens | 279 (0.8%) | 3 (1.1%) |
| twenties | Twenties | 10,340 (28.4%) | 34 (12.0%) |
| thirties | Thirties | 4,754 (13.1%) | 29 (10.2%) |
| fourties | Fourties | 2,158 (5.9%) | 16 (5.7%) |
| fifties | Fifties | 2,809 (7.7%) | 13 (4.6%) |
| sixties | Sixties | 4,376 (12.0%) | 9 (3.2%) |
| seventies | Seventies | 2,159 (5.9%) | 7 (2.5%) |
| eighties | Eighties | - | - |
| nineties | Nineties | - | - |
| - | Unspecified | 9,502 (26.1%) | 192 (67.8%) |
Age declared: 26,875 of 36,377 clips (73.9%), 91 of 283 speakers (32.2%)
Clip buckets
| Bucket | Clips |
|---|---|
| Validated | 33,940 (93.3%) |
| Invalidated | 2,406 (6.6%) |
| Other | 31 (0.1%) |
Training splits
| Split | Clips |
|---|---|
| Train | 4,428 (13.0%) |
| Dev | 3,489 (10.3%) |
| Test | 3,492 (10.3%) |
Training split coverage: 11,409 of 33,940 validated clips (33.6%)
The dataset contains 33940 validated, 2406 invalidated, and 31 unresolved clips. The average clip duration is 3.256 seconds.
Validated sentences: 23,286
| Category | Count |
|---|---|
| Unvalidated sentences | 18,484 |
| Pending sentences | 18,401 |
| Rejected sentences | 83 |
| Reported sentences | 303 |
The corpus contains 41,770 sentences: 23,286 validated and 18,484 unvalidated (18,401 pending review, 83 rejected), with 303 reported for review.
The writing system in this dataset uses general Latin script. The diacritics à, è and é may appear in the transcription of Gwenedeg variant.
a â b ch c'h d e ê f g h i j k l m n ñ o ô p r s t u ù û ü v w y z
à è é (for gwenedeg variant)
c q x (foreign, loanwords)
There follows a randomly selected sample of five sentences from the corpus.
Hanternoz eo.
Buhez a zo ennañ.
Merzhet en deus e oa toull e chakot.
Penaos oc'h kroget ha perak ?
Tu bennak e tleo bezañ.
| Source | Sentences |
|---|---|
| ofis_publik_ar_brezhoneg | 6,506 (27.9%) |
| Desketa (OPAB) | 4,429 (19.0%) |
| OPAB | 4,238 (18.2%) |
| kelaouenn Ya! | 2,679 (11.5%) |
| Grand dictionnaire Français-breton, François Vallée, 1931 | 689 (3.0%) |
| Emveneg | 608 (2.6%) |
| ofis_publik_ar_brezhoneg_parallel | 604 (2.6%) |
| Bilzig, Fañch al Lay, 1925 | 551 (2.4%) |
| An Ti Satanazet, Jakez Riou, 1947 | 530 (2.3%) |
| Other | 2,452 (10.5%) |
| Code | Domain | Clips | Speakers |
|---|---|---|---|
| general | General | 4,496 (12.4%) | 82 (29.0%) |
| agriculture_food | Agriculture and Food | 3 (0.0%) | 3 (1.1%) |
| automotive_transport | Automotive and Transport | 2 (0.0%) | 2 (0.7%) |
| finance | Finance | 4 (0.0%) | 3 (1.1%) |
| service_retail | Service and Retail | 4 (0.0%) | 3 (1.1%) |
| healthcare | Healthcare | 13 (0.0%) | 7 (2.5%) |
| history_law_government | History, Law and Government | 11 (0.0%) | 8 (2.8%) |
| media_entertainment | Media and Entertainment | 13 (0.0%) | 8 (2.8%) |
| nature_environment | Nature and Environment | 8 (0.0%) | 5 (1.8%) |
| news_current_affairs | News and Current Affairs | - | - |
| technology_robotics | Technology and Robotics | 4 (0.0%) | 4 (1.4%) |
| language_fundamentals | Language Fundamentals | - | - |
Each row of a tsv file represents a single audio clip, and contains the following information:
client_id - hashed UUID of a given user
path - relative path of the audio file
text - supposed transcription of the audio
up_votes - number of people who said audio matches the text
down_votes - number of people who said audio does not match text
age - age of the speaker1
gender - gender of the speaker1
accents - accents of the speaker1
variant - variant of the language1
segment - if sentence belongs to a custom dataset segment, it will be listed here
prompt_upvotes - number of upvotes the sentence prompt received
prompt_reports - number of reports the sentence prompt received
is_edited - whether the clip's transcription has been edited
validated_sentences.tsvThe validated_sentences.tsv file contains one row per validated sentence in the text corpus:
sentence_id - unique identifier for the sentence
sentence - the sentence text
variant - the variant of the language
sentence_domain - the domain(s) the sentence belongs to
source - the source the sentence was collected from
is_used - whether the sentence is still in circulation for recording
clips_count - number of clips recorded for this sentence
unvalidated_sentences.tsvThe unvalidated_sentences.tsv file contains one row per unvalidated sentence in the text corpus:
sentence_id - unique identifier for the sentence
sentence - the sentence text
variant - the variant of the language
sentence_domain - the domain(s) the sentence belongs to
source - the source the sentence was collected from
up_votes - number of upvotes the sentence received
down_votes - number of downvotes the sentence received
status - current status of the sentence (pending or rejected)
Ofis Publik ar Brezhoneg (public institution in charge of promoting and developing the teaching and use of the Breton language in everyday life)
Porched niverel ar brezhoneg (portal gathering various digital tools relating to the Breton language)
This dataset is released under the Creative Commons Zero (CC-0) licence. By downloading this data you agree to not determine the identity of speakers in the dataset.
For a full list of age, gender, and accent options, see the demographics spec. These will only be reported if the speaker opted in to provide that information. ↩ ↩2 ↩3 ↩4