Skip to content

Dataset · twangodev

Multilingual LibriVox Audiobooks Mirror

twangodev/librivox-mirror

Over 133,000 hours of voiced books across 86 languages for training speech recognition and synthesis systems.

What's inside

A regularly updated mirror of LibriVox audiobooks, including 21,857 books (496,308 fragments) with a total duration of 133,498.2 hours. Data is presented in audio, text, and tabular formats (100,000 to 1 million records) spanning 86 languages, led by English (132,548.9 hours) alongside German, Spanish, French, and others.

What you can do with it

  • Training and fine-tuning speech recognition (ASR) models
  • Training multilingual text-to-speech (TTS) systems
  • Creating audio and text alignment models
Tasks
Automatic speech recognition, Text → speech
Languages
Multilingual
Size
100K–1M rows
Downloads this month
23K
Likes
0

How to load it

from datasets import load_dataset

ds = load_dataset("twangodev/librivox-mirror")