Semi-Supervision in ASR: Sequential MixMatch and Factorized TTS-Based Augmentation
By Zhehuai Chen, Andrew Rosenberg, Yu Zhang, Heiga Zen, Mohammad Ghodsi, Yinghui Huang, Jesse Emond, Gary Wang, Bhuvana Ramabhadran, Pedro J. Moreno Mengibar. 2021
17 citations recorded by the source corpus.
Semi and self-supervised training techniques have the potential to improve performance of speech recognition systems without additional transcribed speech data. In this work, we demonstrate the efficacy of two approaches to semi-supervision for automated speech recognition. The two approaches leverage vast amounts of available unspoken text and untranscribed audio. First, we present factorized multilingual speech synthesis to improve data augmentation on unspoken text. Next, we propose the Sequential MixMatch algorithm with iterative learning to learn from untranscribed speech. The algorithm is built on top of our online implementation of Noisy Student Training. We demonstrate the compatibility of these techniques yielding an overall relative reduction of word error rate of up to 14.4% on the voice search tasks on 4 Indic languages.
Subfields: Artificial Intelligence, Organic Chemistry, Speech & Audio
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
In Speech & Audio
Citations in the corpus
- BigSSL: Exploring the Frontier of Large-Scale Semi-Supervised Learning for Automatic Speech Recognition (2021, methodology): “There is extensive literature on pretraining [1]–[13] and self-training [14]–[25] in ASR, a subset of which we list in the bibliography.”
- Injecting Text in Self-Supervised Speech Pretraining (2021, background): “Recent advances in semi-supervised learning have revisited unsupervised learning in the form of Noisy Student Training (NST) [9, 10] and introduced augmentation strategies such as FixMatch [11, 12] and Sequential MixMatch [13] to ASR.”