|
--- |
|
dataset_info: |
|
features: |
|
- name: audio |
|
dtype: audio |
|
- name: id |
|
dtype: int64 |
|
- name: time |
|
dtype: string |
|
- name: sentence |
|
dtype: string |
|
- name: orig_file_name |
|
dtype: string |
|
splits: |
|
- name: train |
|
num_bytes: 1440415495.732 |
|
num_examples: 19588 |
|
- name: validation |
|
num_bytes: 169864351.65 |
|
num_examples: 2449 |
|
- name: test |
|
num_bytes: 175082366.835 |
|
num_examples: 2449 |
|
download_size: 1551213354 |
|
dataset_size: 1785362214.2170002 |
|
configs: |
|
- config_name: default |
|
data_files: |
|
- split: train |
|
path: data/train-* |
|
- split: validation |
|
path: data/validation-* |
|
- split: test |
|
path: data/test-* |
|
task_categories: |
|
- automatic-speech-recognition |
|
- conversational |
|
language: |
|
- es |
|
pretty_name: AMERESCO-ASR Subset |
|
size_categories: |
|
- 10K<n<100K |
|
--- |
|
|
|
# Introduction |
|
The "Ameresco-ASR" Subset is a specialized extract from the "Corpus Ameresco" (Albelda and Estellés, online), focusing on colloquial Spanish spoken in various cities across Spain and the Americas. This dataset has been specifically curated to facilitate the fine-tuning of Whisper, an automatic speech recognition system. To achieve this, audio and text segments ranging from 3 to 30 seconds have been automatically extracted from the Ameresco corpus, offering diverse samples of colloquial Spanish from different sociolects and regions. To ensure manageability and efficient processing, a maximum of 1024 tokens were used in the dataset, maintaining a balance between comprehensive coverage and computational efficiency. |
|
|
|
# Content and Geographic Focus |
|
The original Ameresco corpus was initiated as a collaborative project led by Antonio Briz, focusing on the study of colloquial Spanish in European and American geolects. The project included initiatives such as incorporating American usages and Americanisms into the Dictionary of Discourse Particles of Spanish (www.dpde.es), directed by Antonio Briz, Salvador Pons, and José Portolés, as well as the study of attenuation in various Spanish dialects (Projects Es. Var. Atenuación [IP. Marta Albelda], Es VaG. Atenuación [IP Marta Albelda, Maria Estellés]). The project Es.Por.Atenuación, led by Antonio Briz, was also part of this effort, which involved studying attenuation in Portuguese and comparing it with Spanish. Currently, the project is funded by the Esprint project of the Ministry of Science and Innovation (PID2020-114805GB-100, IP Marta Albelda, Maria Estellés). |
|
|
|
The primary outcome of the Ameresco project is the compilation of the Ameresco corpus, which aims to gather samples of colloquial conversations from major cities in Spain, including Santiago de Chile, Tegucigalpa, Temuco, Tucumán, Barranquilla, Buenos Aires, Ciudad de México, Ciudad de Panamá, Iquique, La Habana, Las Palmas, Loja, Medellín, Monterrey, Querétaro, and Santa Cruz in the Americas. This corpus provides a rich resource for studying colloquial Spanish across different regions and sociolects. |
|
|
|
# Transcription approach |
|
|
|
See work documents on https://esvaratenuacion.es/protocolo-de-trabajo |
|
|
|
# References |
|
|
|
Albelda, M. y Estellés, M. (coords.): Corpus Ameresco, Universitat de València, ISSN: 2659-8337, www.corpusameresco.com |
