VoiceInteraction To Showcase Automatic Closed Captioning At NAB
VoiceInteraction will be showcasing the sixth generation of its automatic closed captioning product, Audimus.Media, at next week’s NAB Show (booth SU11021).
The company’s flagship product, Audimus.Media, was built upon VoiceInteraction’s speech processing technology as a proprietary automatic closed captioning solution comprising state-of-the-art signal processing and machine learning techniques for live shows.
It covers captioning accessibility requirements to live TV broadcast, Internet streaming, VOD publishing and video metadata indexing, with advanced audio signal analysis, large vocabulary speech recognition, speaker identification, punctuation and automatic content segmentation to produce highly accurate captions with low latency.
To make sure the language knowledge base is kept up-to-date, alongside crawling online text sources, news scripts are pulled from TV station’s newsroom computer system and feed into machine learning processes to allow constant adaptation to local terms, names and expressions. To cope with speaker accent diversity, alternative word pronunciations are automatically identified and used to widen the accepted range.
Generating automatic closed captions for live TV programs was the starting point but with the viewers moving to live streaming platforms, VoiceInteraction says it can now attach captions directly into video streams ensuring they’ll go all the way along the distribution pipeline. “And since most TV stations distribute video segments over VOD platforms as well, Audimus.Media is now able to automatically export video clips with aligned captions alongside with valuable metadata useful for content indexing and retrieval or search engine optimization.”