BROADCAST RECEIVING DEVICE AND VOICE DICTIONARY CONSTRUCTION PROCESSING METHOD

摘要

A character recognition section (110) recognises a subtitle character sequence displayed on screen from subtitle data contained in a broadcast signal. A voice recognition section (151) acquires voice data output in the period in which the subtitle sequence is displayed on the screen and recognises this as subtitle voice data corresponding to the reading of the subtitle character sequence. An unregistered word extraction section (152) finds subtitle words by analysing the subtitle character sequence into word units, and finds word voice data by analysing the subtitle voice data into word units. The unregistered word extraction section (152) then extracts, as unregistered words, subtitle words whose word voice data is not registered in a voice dictionary (112), and thereby generates a registration candidate list comprising such unregistered words. Based on this registration candidate list, a word registration section (153) registers unregistered words in the voice dictionary (112) together with the corresponding word voice data.