Libri-Light: A Benchmark for ASR with Limited or No Supervision

Jacob Kahn, Morgane Rivière, Weiyi Zheng, Evgeny Kharitonov, Qiantong Xu +10 more
12/17/2019
cs.CLcs.SDeess.AS

Official code: facebookresearch/libri-light (523 stars, MIT, last commit 2023-07-11).

Abstract

We introduce a new collection of spoken English audio suitable for training speech recognition systems under limited or no supervision. It is derived from open-source audio books from the LibriVox project. It contains over 60K hours of audio, which is, to our knowledge, the largest freely-available corpus of speech. The audio has been segmented using voice activity detection and is tagged with SNR, speaker ID and genre descriptions. Additionally, we provide baseline systems and evaluation metrics working under three settings: (1) the zero resource/unsupervised setting (ABX), (2) the semi-supervised setting (PER, CER) and (3) the distant supervision setting (WER). Settings (2) and (3) use limited textual resources (10 minutes to 10 hours) aligned with the speech. Setting (3) uses large amounts of unaligned text. They are evaluated on the standard LibriSpeech dev and test sets for comparison with the supervised state-of-the-art.

View on arXivView PDF

Code Implementations(1)

facebookresearch/libri-lightOfficial100%
MIT

dataset for lightly supervised training using the librivox audio book recordings. https://librivox.org/.

52380Shell, CythonNov 26, 20193 years agoMIT

Similar Papers

GigaSpeech: An Evolving, Multi-domain ASR Corpus with 10,000 Hours of Transcribed Audio

86% match1 repo7312021

Ramsa: A Large Sociolinguistically Rich Emirati Arabic Speech Corpus for ASR and TTS

85% match2026

BanglaKontho: Closing the Long-Form Gap in Bangla Text-to-Speech

84% match2026

FunASR: A Fundamental End-to-End Speech Recognition Toolkit

84% match1 repo19.4k2023

Bootstrapping Audiovisual Speech Recognition in Zero-AV-Resource Scenarios with Synthetic Visual Data

83% match2026

AfriVoices-KE: A Multilingual Speech Dataset for Kenyan Languages

83% match2026

Cite this paper

@article{kahn2019librilight,
  title  = {Libri-Light: A Benchmark for ASR with Limited or No Supervision},
  author = {Jacob Kahn and Morgane Rivière and Weiyi Zheng and Evgeny Kharitonov and Qiantong Xu and Pierre-Emmanuel Mazaré and Julien Karadayi and Vitaliy Liptchinsky and Ronan Collobert and Christian Fuegen and Tatiana Likhomanenko and Gabriel Synnaeve and Armand Joulin and Abdelrahman Mohamed and Emmanuel Dupoux},
  year   = {2019},
  eprint = {1912.07875},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/1912.07875}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/1912.07875.svg)](https://codeofpaper.com/paper/1912.07875)

Discussion