2021 INTERSPEECH INTERSPEECH 2021

Live TV Subtitling Through Respeaking

Abstract

In this paper, we describe our solution for live TV subtitling. The subtitling system uses the respeaking concept with respeakers closely tied with the automatic speech recognition system. The ASR is specially tailored to the live subtitling task by using respeaker-specific acoustic models and TV-show-dependent language models. The output stream of ASR could be online modified by keyboard shortcuts controlled by the respeaker. The whole subtitling service is used by Czech Television to provide high-quality subtitles of live shows for people with hearing impairments.

🧭 Keyword Pioneer — live subtitling
🐝 Cross-Pollinator — Artificial Intelligence, Computer Science, Computer Vision, Data Science & Analytics, Deep Learning, Healthcare & Medicine, Interdisciplinary, Knowledge & Reasoning, Machine Learning, Mathematics & Optimization, Natural Language Processing, Reinforcement Learning, Speech & Audio