Hugging Face's logo

Datasets:
firstpixel
/
pt-br_char

Tasks:
Text-to-Speech
Modalities:
Audio
Languages:
Portuguese
Size:
10K<n<100K
License:
Dataset card Viewer Files Community
pt-br_char / README.md
firstpixel's picture
firstpixel
Update README.md
f7baaff verified
preview code
|
raw
history blame contribute delete
3.21 kB
metadata
license: cc-by-sa-3.0
task_categories:
  - text-to-speech
language:
  - pt
size_categories:
  - 10K<n<100K
configs:
  - config_name: default
    data_files:
      - split: train
        path: train.csv
      - split: metadata
        path: metadata.csv

Brazilian Portuguese Merged Speech Dataset (Derived from Common Voice)

This dataset is a preprocessed and merged version of the Mozilla Common Voice dataset for Brazilian Portuguese (pt-BR). It was created by filtering, merging, and normalizing audio clips to improve usability for speech recognition and TTS (Text-to-Speech) training.

📌 Dataset Details

  • Source: Derived from Common Voice Corpus 20.0
  • Language: 🇧🇷 Brazilian Portuguese (pt-BR)
  • Format: MP3 (24 kHz, mono, 64 kbps)
  • Metadata: metadata.csv file with corresponding transcriptions
  • License: CC BY-SA 3.0

🎯 Preprocessing Steps

This dataset was generated using a Python script that:

  1. Filtered Brazilian Portuguese sentences (removed pt-PT).
  2. Kept only validated recordings (at least one upvote).
  3. Merged short clips from the same speaker (5 to 27 sec per file).
  4. Added 300ms silence between merged sentences.
  5. Normalized the audio:
    • 24 kHz sample rate
    • 16-bit sample width
    • Mono audio
    • 64 kbps bitrate
  6. Generated a transcription file (metadata.csv).

📂 File Structure 

├── wavs/       # Processed audio files (MP3)
│   ├── segment_1.mp3
│   ├── segment_2.mp3
│   ├── ...
├── metadata.csv         # Transcriptions (filename|text)
├── segment_index.json   # Processing progress tracking
├── README.md            # This file

prepare complete samples : 35072 time data : 187:12:22 min sec : 5.06 max sec : 30.0

📜 License & Attribution

This dataset is released under the Creative Commons Attribution-ShareAlike 3.0 (CC BY-SA 3.0) license, the same as the original Common Voice dataset.

To comply with this license:

  • You must give credit to Mozilla Common Voice.
  • If you modify or redistribute this dataset, you must release it under the same CC BY-SA 3.0 license.
  • Do not imply endorsement from Mozilla or Common Voice.

📌 Citation Example: 

Derived from Mozilla Common Voice, licensed under CC BY-SA 3.0.
https://commonvoice.mozilla.org/

🚀 Usage Examples

Python (Load Metadata) 

import pandas as pd

metadata = pd.read_csv("metadata.csv", sep="|")
print(metadata.head())

This is made to be used on F5-TTS, follow instructions from https://github.com/SWivid/F5-TTS/discussions/57 to prepare the dataset.

Using with Speech Models

You can use this dataset for:

  • Training Speech-to-Text (STT) models (e.g., Whisper, DeepSpeech)
  • Building Text-to-Speech (TTS) systems
  • Speaker recognition & voice cloning

💡 Contributing & Feedback

If you have any suggestions, feel free to open an issue or pull request on Hugging Face.