MSR-86K: An Evolving, Multilingual Corpus with 86,300 Hours of Transcribed Audio for Speech Recognition Research

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Li, Song, You, Yongbin, Wang, Xuezhi, Tian, Zhengkun, Ding, Ke, Wan, Guanglu
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866929400657739776
author Li, Song
You, Yongbin
Wang, Xuezhi
Tian, Zhengkun
Ding, Ke
Wan, Guanglu
author_facet Li, Song
You, Yongbin
Wang, Xuezhi
Tian, Zhengkun
Ding, Ke
Wan, Guanglu
contents Recently, multilingual artificial intelligence assistants, exemplified by ChatGPT, have gained immense popularity. As a crucial gateway to human-computer interaction, multilingual automatic speech recognition (ASR) has also garnered significant attention, as evidenced by systems like Whisper. However, the proprietary nature of the training data has impeded researchers' efforts to study multilingual ASR. This paper introduces MSR-86K, an evolving, large-scale multilingual corpus for speech recognition research. The corpus is derived from publicly accessible videos on YouTube, comprising 15 languages and a total of 86,300 hours of transcribed ASR data. We also introduce how to use the MSR-86K corpus and other open-source corpora to train a robust multilingual ASR model that is competitive with Whisper. MSR-86K will be publicly released on HuggingFace, and we believe that such a large corpus will pave new avenues for research in multilingual ASR.
format Preprint
id arxiv_https___arxiv_org_abs_2406_18301
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle MSR-86K: An Evolving, Multilingual Corpus with 86,300 Hours of Transcribed Audio for Speech Recognition Research
Li, Song
You, Yongbin
Wang, Xuezhi
Tian, Zhengkun
Ding, Ke
Wan, Guanglu
Audio and Speech Processing
Computation and Language
Sound
Recently, multilingual artificial intelligence assistants, exemplified by ChatGPT, have gained immense popularity. As a crucial gateway to human-computer interaction, multilingual automatic speech recognition (ASR) has also garnered significant attention, as evidenced by systems like Whisper. However, the proprietary nature of the training data has impeded researchers' efforts to study multilingual ASR. This paper introduces MSR-86K, an evolving, large-scale multilingual corpus for speech recognition research. The corpus is derived from publicly accessible videos on YouTube, comprising 15 languages and a total of 86,300 hours of transcribed ASR data. We also introduce how to use the MSR-86K corpus and other open-source corpora to train a robust multilingual ASR model that is competitive with Whisper. MSR-86K will be publicly released on HuggingFace, and we believe that such a large corpus will pave new avenues for research in multilingual ASR.
title MSR-86K: An Evolving, Multilingual Corpus with 86,300 Hours of Transcribed Audio for Speech Recognition Research
topic Audio and Speech Processing
Computation and Language
Sound
url https://arxiv.org/abs/2406.18301