Brazilian Portuguese Voice Dataset (brptvl)

A Brazilian Portuguese (PT-BR) voice dataset containing speech audio recordings for Automatic Speech Recognition (ASR) and voice-related applications.

Dataset Description

This dataset provides Brazilian Portuguese speech audio collected for voice and speech recognition purposes. The data is stored in two compressed archives and is currently set to private visibility on the Hugging Face Hub.

Use Cases

• Brazilian Portuguese ASR: Train or fine-tune speech-to-text models for the Brazilian Portuguese language
• Voice Biometrics: Develop speaker verification and identification systems for PT-BR speakers
• Text-to-Speech (TTS): Source voice recordings for synthesizing Brazilian Portuguese speech
• Language Adaptation: Adapt existing multilingual models to the Brazilian Portuguese variant

File Inventory

File	Description	Size
valentino.zip	Brazilian Portuguese audio recordings (first batch)	~0.8 GB
vlbrpt.zip	Brazilian Portuguese audio recordings (second batch)	~0.8 GB
.gitattributes	Git LFS configuration for large file storage	—
README.md	Dataset documentation (this file)	—

Usage

Loading the Dataset

from datasets import load_dataset

dataset = load_dataset("edwixx/brptvl", trust_remote_code=True)

Extracting Audio

# Extract audio archives
unzip valentino.zip -d valentino_audio/
unzip vlbrpt.zip -d vlbrpt_audio/

Access Note

This dataset is private. Only the repository owner and authorized collaborators can access it. To request access, contact the dataset author.

Metadata

• Author: Anurag Kanade
• Language: Brazilian Portuguese (pt-BR)
• License: MIT
• Hosted by: Hugging Face Datasets
• Visibility: Private
• Created: 2025-06-03
• Last Updated: 2026-06-25
• Storage Size: ~1.7 GB
• Downloads: 0

Citation

@misc{edwixx-brptvl,
  author = {Anurag Kanade},
  title = {brptvl},
  year = {2026},
  publisher = {Hugging Face},
  journal = {Hugging Face Hub},
  howpublished = {\url{https://huggingface.co/datasets/edwixx/brptvl}}
}