European Commission Joint Research Centre, Institute for the Protection and Security of the Citizen, VOICE Project

English

Home
Objectives
Events
Education
Media
VoiceLab
Forum
Path:
Objectives > Annual Reports > ISIS Success Stories

VOICE Project's Objectives
ISIS Success Stories

PUBLICATION:

LANG.:

SUBJECT:

LENGTH:

STORY NO:

JRC STORIES

EN

VOICE PROJECT

701 WORDS

JRC-31-VOI

A significant number of Europeans suffer hearing difficulties. They gain no benefit from the telephone, non-subtitled television broadcasts or oral presentations. The European Commission's Joint Research Centre is encouraging the development of new equipment and in raising awareness among the hearing population, which will help tackle the problems. Advances include better subtitling software and voice-to-text systems for telephony.

Up to 5% of Europeans suffer from chronic total or partial deafness. This does not include hearing problems in the elderly, who form an ever-increasing proportion of the community. Those with no hearing problems benefit from the communications revolution with voice telephony taken for granted and easy access to vast amounts of television broadcasts. But a telephone is not a good medium for someone who cannot hear. And, although some television broadcasts are subtitled, most are not. The same is true for conferences as well as school and university lessons. Europeans with hearing problems are being excluded.

Man and machine

The partners in the VOICE project - co-ordinated by the Joint Research Centre's Institute for Systems, Informatics and Safety (ISIS) in Ispra, Italy - aim to correct the situation by meeting two main challenges. The first is developing software interfaces for adapting inexpensive 'voice to text' (VTT) systems, which will allow the spoken, or digitally recorded, word to be translated into text that can be read on a PC's screen. The second, probably more important, is to raise awareness of the problems faced by the hearing impaired. Addressing the issue of awareness, partners in the consortium are encouraging interested parties to communicate more effectively by setting up a comprehensive web site (http://voice.jrc.it). The site has various interactive elements. And a 'special interest user group' defines user requirements and airs their views in a users' VOICE 'forum'. The team is also holding a programme of conferences with participants from industry, academia and broadcasting considering the problems faced by the hearing impaired. It is here that the technical and human sides of the project overlap. One initial result is an interface which allows IBM or Dragon speech-recognition software to be used to subtitle presentations. These programmes were developed originally to produce text for use in word processing. On their own, they are not idea for producing subtitles. The screen is rather cluttered, the fonts quite small and the speaker has to dictate punctuation marks. The new interface produces a display where the recognised text is transformed into subtitles, much easier to read. The human involvement does not stop there. The speakers get the full control of the visual communication channel. They must modify their style of presentation, meeting the needs of persons with hearing impairment. They have to speak a little slower and pause between phrases, thus controlling the subtitles' length on the screen. When they do, the subtitling is up to 95% accurate and the audience will have the time to read the subtitles.

Reading the phone

Voice telephony is also being investigated. Current VTT systems require 'training' in recognising the individual user's voice, accent and pronunciation before they can produce an accurate text. Over the telephone line the accuracy is reduced by noise and bandwidth limits. While some software will recognise virtually anyone's voice, the vocabulary is limited to around 30 words. As an interim step, VTT systems are being developed that could recognise a vocabulary of up to a thousand words from a limited number of people. In this case, the deaf person's immediate family could communicate fairly effectively. The people involved would need a few hours training.

Improving quality of life

The project is long-term but worthwhile. It fits well in the Fifth Framework Programme's themes of access to the information society and improving quality of life by trying to include a portion of the population that is currently, at least partially, excluded. Thanks to a 'design for all' approach, anyone will benefit from this advance. Subtitling of conferences, school lessons, television broadcasts and video-telephone calls will facilitate the integration of persons with special needs, including the elderly and the immigrated, and of anyone listening a language different from his mother tongue.


Map
Accessibility
Copyright
Contacts
top of page