CORAA-MUPE-ASR / README.md
sidleal's picture
Update README.md
d3d5f9c verified
metadata
dataset_info:
  features:
    - name: audio_id
      dtype: int64
    - name: audio_name
      dtype: string
    - name: file_path
      dtype: string
    - name: speaker_type
      dtype: string
    - name: speaker_code
      dtype: string
    - name: speaker_gender
      dtype: string
    - name: education
      dtype: string
    - name: birth_state
      dtype: string
    - name: birth_country
      dtype: string
    - name: age
      dtype: int64
    - name: recording_year
      dtype: int64
    - name: audio_quality
      dtype: string
    - name: start_time
      dtype: float32
    - name: end_time
      dtype: float32
    - name: duration
      dtype: float32
    - name: normalized_text
      dtype: string
    - name: original_text
      dtype: string
    - name: audio
      dtype:
        audio:
          sampling_rate: 16000
    - name: racial_category
      dtype: string
  splits:
    - name: validation
      num_bytes: 1414790171.812
      num_examples: 9894
    - name: test
      num_bytes: 3801921623.48
      num_examples: 30968
    - name: train
      num_bytes: 36963919315.24
      num_examples: 276881
  download_size: 41830719092
  dataset_size: 42180631110.532
configs:
  - config_name: default
    data_files:
      - split: validation
        path: data/validation-*
      - split: test
        path: data/test-*
      - split: train
        path: data/train-*

MuPe Life Stories Dataset

A new publicly available dataset consisting of 289 life story interviews (365 hours), featuring a broad range of speakers varying in age, education, and regional accents.

Metadata:

  • audio_id: Sequential id for the interview;
  • audio_name: Unique code for the interview;
  • file_path: Wav audio file path;
  • speaker_type: R is the interviewee, P/1 is interviewer 1, P/2 is interviewer 2 and so on;
  • speaker_code: Unique code for the speaker;
  • speaker_gender: Gender of the speaker;
  • education: Education level of the interviewee, filled only when speaker_type = 'R';
  • birth_state: Birth state (region) of the interviewee, filled only when speaker_type = 'R';
  • birth_country: Birth country of the speaker;
  • age: Age of the interviewee, calculated with recording_year minus year of birth, filled only when speaker_type = 'R';
  • recording_year: The year when the audio was recorded;
  • audio_quality: Can be high or low;
  • start_time: The start time in the original complete audio file;
  • end_time: The end time in the original complete audio file;
  • duration: The duration of the segment;
  • normalized_text: Text normalized in lowercase and without punctuation marks;
  • original_text: Text before normalization.
  • racial_category: Racial category may be Black, White, Pardo (Mixed) and Asian. Few interviews have this information.

Dataset

Model

Citation

Leal, S.E.; Candido Junior, A.; Marcacini, R.; Casanova, E.; Gonçalves, O.; Soares, A.; Lima, R.; Gris, L.; Aluísio, S.M. MuPe Life Stories Dataset: Spontaneous Speech in Brazilian Portuguese with a Case Study Evaluation on ASR Bias against Speakers Groups and Topic Modeling. Proceedings of the 31st International Conference on Computational Linguistics (COLING) (2025).

@inProceedings{Leal2025Coling,
author={Sidney Leal
   and Arnaldo Candido Jr.
   and Ricardo Marcacini
   and Edresson Casanova
   and Odilon Gonçalves
   and Anderson Soares
   and Rodrigo Lima
   and Lucas Gris
   and Sandra Alu{\'i}sio,
title={MuPe Life Stories Dataset: Spontaneous Speech in Brazilian Portuguese with a Case Study Evaluation on ASR Bias against Speakers Groups and Topic Modeling},
booktitle={Proceedings of the 31st International Conference on Computational Linguistics (COLING)},
year={2025}
}

Sponsors / Funding

This work was carried out at the Center for Artificial Intelligence (C4AI-USP), with support by the São Paulo Research Foundation (FAPESP grant #2019/07665-4) and by the IBM Corporation. This project was also supported by the Ministry of Science, Technology and Innovation, with resources of Law No. 8.248, of October 23, 1991, within the scope of PPI-SOFTEX, coordinated by Softex and published Residence in TIC 13, DOU 01245.010222/2022-44. This work has been partially supported by Advanced Knowledge Center in Immersive Technologies (AKCIT/CEIA), with financial resources from the PPI IoT/Manufatura 4.0 / PPI HardwareBR of the MCTI, signed with EMBRAPII.