Spoken subtitles – if, how, for whom and why?


The article describes a system for automatic reading and broadcasting subtitles, based on synthetic speech files generated from TV subtitle files and transmitted to the television ether through a separate audio channel. Owing to the recent progress in the development of speech technology (speech synthesis and speech recognition) and the growing popularity of digital television, people with special linguistic needs such as those suffering from visual or reading impairments, can access various services connected with mass media. The software solution of spoken subtitles consists of two logical components, namely, an interface of automatic reading and voice editing and another interface for the management and transmission of subtitle and speech files. The editor interface analyses the subtitles and makes a list of those foreign names whose pronunciation is not available in the foreign names database, enabling the editor to add pronunciations and check the correctness of the spoken subtitles. In addition, the editor can, in principle, select among different synthetic voices and change their parameters, if necessary. An Estonian speech synthesizer translates subtitles into speech and synchronizes it with the film or broadcast. The ether interface manages and controls the data streams for automatic reading and is responsible for mixing the spoken subtitle files and the original broadcast sound as well as for their transmission into the ether. The spoken subtitles are, first and foremost, meant to help people with reading impairment and those of the visually challenged who can see what is going on in the TV screen, but cannot read the small font of the subtitles. Besides the dyslectics and the visually handicapped, spoken subtitles could be of help to the elderly TV-viewers whose vision has dropped, and to the children who cannot read really well yet. Also, students of Estonian as a foreign language could be interested in the service. Although at present the subtitles read by synthetic voice and transmitted through a separate sound channel can be used only with the ETV and ETV2 channels of Estonian Public Broadcasting, the software solution is freeware, available free of charge to all Estonian media companies.