You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this dataset content.

Vietnamese News ASR Dataset

Mô tả Dataset

Dataset tiếng Việt cho bài toán Automatic Speech Recognition (ASR) được thu thập từ các video tin tức trên YouTube.

Thông tin Dataset

  • Ngôn ngữ: Tiếng Việt
  • Tổng số clips: 8,130
  • Tổng thời gian: 36.43 giờ
  • Sample rate: 16kHz
  • Format: WAV

Phân bố dữ liệu

Theo Label

  • Miền Bắc: 3,021 clips
  • Miền Nam: 2,612 clips
  • Miền Trung: 2,497 clips

Train/Test Split

  • Train: 90%
  • Test: 10%

Cấu trúc Dataset

Mỗi sample trong dataset bao gồm:

  • audio: File âm thanh WAV
  • text: Phiên âm (transcription)
  • duration: Độ dài clip (giây)
  • label: Nhãn phân loại (vùng miền/chủ đề)
  • video_id: ID của video nguồn
  • url: Link YouTube nguồn

Sử dụng Dataset

from datasets import load_dataset

# Load dataset
dataset = load_dataset("vinhdq17test/vinhdq17")

# Xem một sample
print(dataset["train"][0])

Downloads last month
5