projecte-aina/commonvoice_benchmark_catalan_accents

Common Voice Benchmark Catalan Accents

Dataset Summary

This is a new presentation of the corpus Catalan Common Voice v17 - metadata annotated version with the splits redefined to benchmark ASR models with various Catalan accents: From the validated recording split, we have selected, for each of the main accents of the language (balearic, central, northern, northwestern, valencian), the necessary male and female speakers to gather approximately two and a half hours of varied voice recordings. Thus, we have created ten benchmarking splits, resulting from the combinations of the 5 accents and 2 genders (female_feminine and male_masculine) taken into account.

The recordings of speakers who have not been included in these splits have been grouped into the training split.

Supported Tasks and Leaderboards

Automatic Speech Recognition.

Languages

The dataset is in Catalan (ca).

Dataset Structure

Data Instances

{
  'client_id': '69dafb41ddc0ea2785719305fdc5c8d79c4b2829d9f3325bda707dcaa553f95c5fbf4b072970d9004d3e31543fcb2c55e252dc904c4fb5aee2a5e5500df90967', 
  'path': 'common_voice_ca_19909748.mp3', 
  'sentence': 'En el carrer de l'església es troben bona part dels edificis importants de la vila.', 
  'up_votes': 2, 
  'down_votes': 0, 
  'age': 'thirties', 
  'gender': 'male_masculine', 
  'accent': 'balear', 
  'variant': '',
  'locale': 'ca', 
  'segment': '',
  'mean quality': '4.0',
  'stdev quality': '0.0',
  'annotated_accent': 'balearic',
  'annotated_accent_agreement': '100.0',
  'annotated_gender': 'male',
  'annotated_gender_agreement': '100.0',
  'propagated_gender': 'male_masculine',
  'propagated_accents': 'balear',
  'propagated_accents_normalized': 'balearic',
  'assigned_accent': 'balearic',
  'assigned_gender': 'male_masculine'
  
}

Data Fields

Data Fields are kept from Catalan Common Voice v17 - metadata annotated version.

Please refer to the README for a detailed explanation of the annotations.

Most of the data fields come from the original Common Voice corpus:

• client_id(string): An id for which client (voice) made the recording

• path(string): The path to the audio file

• sentence_id(string): An id for the text sentence

• sentence(string): The sentence the user was prompted to speak

• sentence_domain(string): Semantic domain of the sentence

• up_votes(int64): How many upvotes the audio file has received from reviewers

• down_votes(int64): How many downvotes the audio file has received from reviewers

• age(string): Self-reported age of the speaker (e.g. teens, twenties, fifties)

• gender(string): Self-reported gender of the speaker

• accent(string): Self-reported accent of the speaker

• locale(string): The locale of the speaker

• segment(string): Usually an empty field

In the annotated version of the corpus, we have added the following fields:

• annotated_gender(string): Annotated gender by the experts team.

• annotated_gender_agreement (float): Agreement whithin the annotation team about the gender of the speaker.

• annotated_accent (string): Annotated accent by the experts team. The accents considered are: Balearic, Central, Northern, Northwestern, Valencian.

• annotated_accent_agreement (float): Agreement whithin the annotaion team about the accent of the speaker.

• mean quality (float): Mean annotated quality of the speakers' recording.

• stdev quality (float): Deviation in the quality annotation between annotators.

• propagated_gender (string): Self-declared gender as indicated in certain recordings by the user. Speakers that change self-declared gender have been labeled as "other.

• propagated_accents (string): Self-declared accent as indicated in certain recordings by the user. See annotations for more information.

• propagated_accents_normalized (string): Propagated accent, normalized to the closed-options list used until version 7.

• assigned_accent (string): Accent assigned to the speaker.

• assigned_gender (string): Gender assigned to the speaker.

Data Splits

The splits have been reworked to obtain two and a half hours for each of the combinations of the 5 accents and 2 genders considered.

split	sentences	speakers	duration (ms)	duration (h)
balearic_female.tsv	1665	131	9066912	2.52
balearic_male.tsv	1616	112	9129120	2.54
central_female.tsv	1742	301	9028276	2.51
central_male.tsv	1701	342	9011986	2.50
northern_female.tsv	1627	55	9402612	2.61
northern_male.tsv	1615	68	9249720	2.57
northwestern_female.tsv	1618	120	9136129	2.54
northwestern_male.tsv	1626	133	9055302	2.51
train.tsv	1801369	32894	9730691599	2702.97
valencian_female.tsv	1744	119	9107568	2.53
valencian_male.tsv	1631	151	9003500	2.50

Dataset Creation

Curation Rationale

In light of the lack of data in diverse accents to evaluate Catalan ASR models, we have reworked the data from the Catalan Common Voice v17 - metadata annotated version to create a benchmark dataset.

We hope that this corpus serves to provide access to speech technologies for Catalan speakers, a minority language, in all its accents.

Source Data

Initial Data Collection and Normalization

The original data comes from Catalan Common Voice v17 - metadata annotated version.

Please refer to the README file for a detailed explanation of the annotations.

Who are the source language producers?

The Common Voice project is an initiative of the Mozilla Foundation to collect voices in various languages and accents. The voices have been provided by volunteers.

For more information, visit the project website.

Annotations

Annotation process

To make this benchmark we have used the annotations from Catalan Common Voice v17 - metadata annotated version.

Detailed information on the annotation process can be found in the README file of the dataset.

Who are the annotators?

The datset Catalan Common Voice v17 - metadata annotated version contains the Common Voice project's own annotations and some annotations made by a team of experts from the UB (University of Barcelona). For detailed information on the process, see the README file.

Personal and Sensitive Information

The dataset consists of people who have donated their voice online. You agree to not attempt to determine the identity of speakers in the Common Voice dataset.

Considerations for Using the Data

Social Impact of Dataset

The dataset consists of people who have donated their voice online. You agree to not attempt to determine the identity of speakers in the Common Voice dataset.

We hope that this corpus serves to provide access to speech technologies for Catalan speakers, a minority language, in all its accents.

Discussion of Biases

Most of the voices of the common voice in Catalan correspond to men with a central accent between 40 and 60 years old. We have reworked the data from the Common Voice in Catalan to create an ASR benchmark dataset considering the most habitual genders (female and male) and accents (balearic, central, northern, northwestern, valencian).

The benchmark does not currently evaluate other genres and accents due to the lack of data. We hope to expand it later.

Regarding the content of the recorded sentences, we consider that the Common Voice validation system is efficient in removing those that could produce toxic content.

Other Known Limitations

[N/A]

Additional Information

Dataset Curators

Language Technologies Unit at the Barcelona Supercomputing Center ([email protected])

This work is funded by the Ministerio para la Transformación Digital y de la Función Pública - Funded by EU – NextGenerationEU within the framework of the project ILENIA with reference 2022/TL22/00215337.

Licensing Information

This dataset can be used for any purpose, whether academic or commercial, under the terms of the CC BY 4.0. Give appropriate credit, provide a link to the license, and indicate if changes were made.

Citation Information

DOI []

Contributions

The manual annotation of the Catalan Common Voice v17 - metadata annotated version was entrusted to the STeL team from the University of Barcelona.