Weakly Supervised Construction of ASR Systems from Massive Video Data
Mengli Cheng, Chengyu Wang, Jun Huang, Xiaobo Wang · 2021
Building large-scale Automatic Speech Recognition (ASR) systems from scratch is significantly challenging, mostly due to the time-consuming and financially-expensive process of annotating a large amount of audio data with transcripts.Although several unsupervised pre-training models have been proposed, applying such models directly might be sub-optimal if more labeled, training data could be obtained without a large cost.In this paper, we present a weakly supervised framework for constructing ASR systems with massive video data.As videos often contain human-speech audio aligned with subtitles, we consider videos as an important knowledge source, and propose an effective approach to extract high-quality audio aligned with transcripts from videos based on text detection and Optical Character Recognition.The underlying ASR model can be fine-tuned to fit any domain-specific target training datasets after weakly supervised pre-training.Extensive experiments show that our framework can easily produce state-of-the-art results on six public datasets for Mandarin speech recognition.