Common Voice Scripted Speech 25.0 - Swahili

License:

CC0-1.0

Steward:

Common Voice

Task: ASR

Release Date: 3/23/2026

Format: MP3

Size: 20.87 GB

Description

A collection of read speech recordings in Swahili (Kiswahili).

Specifics

Licensing

Creative Commons Zero v1.0 Universal (CC0-1.0)

https://spdx.org/licenses/CC0-1.0.html

Considerations

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.

Processes

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.

Metadata

Kiswahili — Swahili (`sw`)

This datasheet is for cv-corpus-25.0-2026-03-09 of the Mozilla Common Voice Scripted Speech dataset for Swahili [Kiswahili - sw]. The dataset contains 730187 clips representing 1064.02 hours of recorded speech (392.11 hours validated) from 1518 speakers, recorded from a text corpus of 140,486 sentences.

Language

Variants

Code	Variant	Clips	Speakers
sw-baratz	Kiswahili cha Bara ya Tanzania	22,486 (3.1%)	24 (1.6%)
sw-sanifu	Kiswahili Sanifu (EA)	21,833 (3.0%)	69 (4.5%)
sw-barake	Kiswahili cha Bara ya Kenya	6,487 (0.9%)	59 (3.9%)
sw-kingwana	Kingwana (DRC)	1,756 (0.2%)	34 (2.2%)
sw-kimvita	Kimvita (KE) - Central dialect	15 (0.0%)	2 (0.1%)
sw-katanga	Katanga (DRC)	5 (0.0%)	1 (0.1%)

Accents

Code	Accent	Clips	Speakers
-		8,022 (1.1%)	50 (3.3%)

Demographic information

The dataset includes the following self-declared age and gender distributions. A coverage summary is shown below each table.

Gender

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	249,195 (34.1%)	399 (26.3%)
female_feminine	Female, feminine	267,669 (36.7%)	399 (26.3%)
transgender	Transgender	-	-
non-binary	Non-binary	-	-
do_not_wish_to_say	Prefer not to say	-	-
-	Unspecified	213,323 (29.2%)	941 (62.0%)

Gender declared: 516,864 of 730,187 clips (70.8%), 577 of 1,518 speakers (38.0%)

Age

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	3,843 (0.5%)	25 (1.6%)
twenties	Twenties	364,209 (49.9%)	705 (46.4%)
thirties	Thirties	79,508 (10.9%)	78 (5.1%)
fourties	Fourties	25,305 (3.5%)	15 (1.0%)
fifties	Fifties	30,175 (4.1%)	7 (0.5%)
sixties	Sixties	4,444 (0.6%)	3 (0.2%)
seventies	Seventies	-	-
eighties	Eighties	-	-
nineties	Nineties	-	-
-	Unspecified	222,703 (30.5%)	922 (60.7%)

Age declared: 507,484 of 730,187 clips (69.5%), 596 of 1,518 speakers (39.3%)

Data splits for modelling

Clip buckets

Bucket	Clips
Validated	269,087 (36.9%)
Invalidated	80,838 (11.1%)
Other	380,262 (52.1%)

Training splits

Split	Clips
Train	46,735 (17.4%)
Dev	12,269 (4.6%)
Test	12,270 (4.6%)

Training split coverage: 71,274 of 269,087 validated clips (26.5%)

The dataset contains 269087 validated, 80838 invalidated, and 380262 unresolved clips. The average clip duration is 5.246 seconds.

Text corpus

Validated sentences: 134,729

Category	Count
Unvalidated sentences	5,757
Pending sentences	5,727
Rejected sentences	30
Reported sentences	1,899

The corpus contains 140,486 sentences: 134,729 validated and 5,757 unvalidated (5,727 pending review, 30 rejected), with 1,899 reported for review.

Sample

There follows a randomly selected sample of five sentences from the corpus.

Kama uko na mashamba ni salama
mtu huyo alikaaa ndani ya maji kwa muda mrefu
Kabila la maasai ni la watu wanaopatikana Afrika ya mashariki
mmoja haswa katika miji midogo Wanasaidia katika
Kuna mnazi moya wapende vachu wachachu akianguka moya hamadi mwana wavachu

Sources

Source	Sentences
batch-2-07-02-2022	39,636 (29.4%)
Kiswahili MCV writing competition	31,752 (23.6%)
batch-1-19-11-2021-contributors-set	28,589 (21.2%)
wiki	8,833 (6.6%)
batch-4-08-02-2023-contributors-set	8,822 (6.5%)
batch-2-10-03-2023-agriculture-set	4,570 (3.4%)
batch-3-17-08-2022-contributors-set	4,350 (3.2%)
batch-2-11-07-2022-contributors-set	2,305 (1.7%)
batch-1-16-02-2023-badili-livestock-diseases-set	1,627 (1.2%)
Other	4,245 (3.2%)

Fields

Clips

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.tsv`

The 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.tsv`

The 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)

Get involved

Community links

Discussions

Contribute

Licence

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.

Footnotes

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