October 25, 2021
Despite rapid progress in the recent past, current speech recognition systems still require labeled training data which limits this technology to a small fraction of the languages spoken around the globe. This paper describes wav2vec-U, short for wav2vec Unsupervised, a method to train speech recognition models without any labeled data. We leverage self-supervised speech representations to segment unlabeled audio and learn a mapping from these representations to phonemes via adversarial training. The right representations are key to the success of our method. Compared to the best previous unsupervised work, wav2vec-U reduces the phone error rate on the TIMIT benchmark from 26.1 to 11.3. On the larger English Librispeech benchmark, wav2vec-U achieves a word error rate of 5.9 on test-other, rivaling some of the best published systems trained on 960 hours of labeled data from only two years ago. We also experiment on nine other languages, including low-resource languages such as Kyrgyz, Swahili and Tatar. The code will be open sourced.
Publisher
NeurIPS
Research Topics
February 24, 2023
Faisal Azhar, Hugo Touvron, Armand Joulin, Aurelien Rodriguez, Baptiste Rozière, Eric Hambro, Gautier Izacard, Guillaume Lample, Marie-Anne Lachaux, Naman Goyal, Thibaut Lavril, Timothee Lacroix, Xavier Martinet, Edouard Grave
February 24, 2023
February 20, 2023
Maziar Sanjabi, Aaron Chan, Hamed Firooz, Lambert Mathias, Liang Tan, Shaoliang Nie, Xiaochang Peng, Xiang Ren
February 20, 2023
December 31, 2022
Yossef Mordechay Adi, Abdelrahman Mohamed, Adam Polyak, Emmanuel Dupoux, Evgeny Kharitonov, Jade Copet, Morgane Rivière, Tu Anh Nguyen, Wei-Ning Hsu, Felix Kreuk
December 31, 2022
December 29, 2022
Dexter Ju, Jason Weston, Sainbayar Sukhbaatar, Stephen Roller
December 29, 2022