Loading datasets...
Task: ASR
Release Date: 6/17/2026
Format: MP3
Size: 173.39 MB
Share
A collection of spontaneous responses to questions in Toba Qom (tob).
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.
tob)This datasheet is for sps-corpus-4.0-2026-06-12 of the Mozilla Common Voice Spontaneous Speech dataset for Toba Qom [tob - tob]. The dataset contains 1572 clips representing 10 hours of recorded speech (9.67 hours validated) from 25 speakers.
The Toba Qom language is an endangered language spoken in Gran Chaco, a region spanned over Argentina, Paraguay and Bolivia. As per the official demographic data provided by the Argentinian state, the population of Qom individuals is estimated at 80,000, of which approximately 49% are speakers of the oral form of the language. The term "qom" describes a population that has traditionally been arranged into multiple extended families or groups. Language and sociocultural traits that are essential to qom culture are shared by these groups, which are traditionally hunter-gatherer.
The contributors to this corpus originate from Chaco and Formosa provinces in Argentina. This area encompasses four ethnodialectal subregions with distinct self-identification terms (Messineo, 1991) [^3].
| Area | Province | Locations | Variant (self-identification) |
|---|---|---|---|
| Northwest | Chaco | El Colchwón, El Espinillo and the Bermejo river’s surroundings | dapigemlʔek |
| Northcenter | Chaco | Pampa del Indio | noʔolgaGanaq |
| Southcenter | Chaco | Sáenz Peña, Machahay, Quitilipi | lʔañaGashek |
| Southeast | Chaco, Eastern Formosa | Las Palmas, Clorinda | takshek |
For further information, see [^2] [^3] [^4].
The dataset clips are categorised by transcription status and training-set assignment. The following tables summarise the distribution.
| Bucket | Clips | % |
|---|---|---|
| Transcribed & Validated | 1,540 | 98.0% |
| Transcribed & Pending | 0 | 0.0% |
| Not transcribed | 32 | 2.0% |
| Bucket | Clips | % |
|---|---|---|
| Train | 917 | 58.3% |
| Dev | 298 | 19.0% |
| Test | 325 | 20.7% |
| Unassigned | 32 | 2.0% |
Training split coverage: 1,540 of 1,540 transcribed & validated clips (100.0%)
| Bucket | Clips | % |
|---|---|---|
| Validated | 1,540 | 100.0% |
| Pending | 0 | 0.0% |
| Edited | 540 | 35.1% |
The transcriptions follows the orthographic systems proposed by Buckwalter (2001) [^2]
a c ch d e g hu i j l ll m n ñ o p q qu r s sh t u v x y ỹ ’
There follows a randomly selected sample of questions used in the corpus.
¿Neque’t nyigaxanaxac so ’an’aixaatega can ’am cacho’olec?
¿Qonetec da ’aulalec que’eca depaqtaxaalec na shiỹiaxauapi yi ’adma’?
¿Qonetec naxa da mashe qoỹoqtegue da sa anauace’ ca ’anauochaxaua?
¿’Aushet da ’au’ogoxona aca nanguishi da natannaxanaqle ’anqaen?
¿’Eetec ca ’adpetaxa ca naỹalec, yi ’adma’ qataq na shiỹaxaupi huetalec ca ’analhua ’aupa’aỹ’ague?
There follows a randomly selected sample of transcribed responses from the corpus.
*aiem can ñaq seraxatac sayocopita nagara aiem canquetoxon dam tet'o lamaxa da a'loñe iataqta sayoqpita yaye cohioqta añi colloxochegue satapegueua na paxaguenaxaqui na ñaqpioleq iataqta caioqta eda da so ñoqpita da sapeteñe chaaye aca ca cannotaxaqui *
Ayem qo ayem yi nishitaique da saya na iqayañicpi qoyoqta ca hueta'a.
Aýem da huo'o ca ñivacacion nache saýia que'eca ýihuo' ñamaqtega
nahua yoogo sachegotahua aca ca imetac na noxot
*Aiem sasayaten da samaqñigueaca ca notaxaqui t chaye ca aca lalamaqte chochesoxa da can sotaa' na huataxanaq nache aiem iapaxagueneq samaqñigue asa so nalataqnaqte *
To be updated in the next release. Contact the author for details.
Each row of a tsv file represents a single audio clip, and contains the following information:
client_id - hashed UUID of a given user
audio_id - numeric id for audio file
audio_file - audio file name
duration_ms - duration of audio in milliseconds
prompt_id - numeric id for prompt
prompt - question for user
transcription - transcription of the audio response
votes - number of people that who approved a given transcript
age - age of the speaker1
gender - gender of the speaker1
language - language name
split - for data modelling, which subset of the data does this clip pertain to
char_per_sec - how many characters of transcription per second of audio
quality_tags - some automated assessment of the transcription--audio pair, separated by |
transcription-length - character per second under 3 characters per second
speech-rate - characters per second over 30 characters per second
short-audio - audio length under 2 seconds
long-audio - audio length over 5 minutes
non-allowed-script - transcription contains characters from a writing system not associated with the language
mixed-script-words - a single word contains characters from multiple writing systems
mixed-script-transcription - transcription spans multiple writing systems, but each word consistently uses only one
Belu Ticona <mticonao@gmu.edu>
Paola Cúneo
Antonios Anastasopoulos
B. Ticona, P. Cuneo. A. Anastasopoulos. “Datasheet of Spontaneous Speech Corpus for Qom - Mozilla Common Voice”. Revised on Aug 29th, 2025. [Publication Date].
This dataset was partially funded by the Open Multilingual Speech Fund managed by Mozilla Common Voice.
The speaker collaborators were funded by Mozilla Common Voice. The project coordinator was partially funded by the US NSF grants 2346334 and 2439202.
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