from the Web:
http://www.ebu.ch/trev_home.htm
No. 296 (October 2003)
European Broadcasting Union, L'Ancienne Route 17A,
CH-1218 Grand-Saconnex,
Geneva, Switzerland, techreview@ebu.ch
 |
 |
The fact that 2003 is the European Year for Disabled People prompts me to ask "What are broadcasters, and particularly EBU Members, doing to meet the needs of disabled people?"
Radio and TV services are important to almost all members of the public – but they can be even more important to disabled people. Blind people are heavily dependent on radio services, whilst people who are deaf or hard-of-hearing require "subtitles" or "signing" to make sense of TV broadcasts. It may be surprising to learn that more than 90% of blind people regularly "watch" TV programmes. Of course, they only listen to the soundtrack, but some programmes are impossible to follow unless you can see the accompanying pictures. Consequently, there is a need for "audio descriptive services" on TV in which an additional audio channel provides crucial spoken descriptions such as "a huge shark jumps out of the water". After a trial of audio description, one blind viewer is reported to have said "Normally after the first ten minutes of a film, I give up. Without the [audio] description, I wouldn't have been able to watch it."
Broadcasters can dramatically improve the lives of disabled people by enhancing TV programmes by providing "access services", namely:
Provision of these "access services" raises several difficult problems, some of which are technical whilst others are economic.
Subtitling is widely used on analogue TV services: for example, 6% of the UK population use subtitling frequently and 13% use it occasionally. Subtitling is successful because it is simple to use and it does not require specialised receiving equipment (all modern TV sets have teletext).
Many broadcasters now offer subtitling on a substantial proportion of their programmes (live and pre-recorded). It is obviously a significant challenge to offer subtitling of live unscripted programmes, but the quality of live subtitling has improved in recent years. Most live subtitling is performed by expert operators using special equipment, such as Palantype or Stenograph machines. Some broadcasters (such as the BBC in the UK and DR in Denmark) have successfully generated subtitles using speech recognition software. In both cases, the words are re-voiced by a human "translator" before being delivered to the speech recognition software.
This approach offers much higher accuracy than attempting a direct transcription of the broadcast audio signal, because (i) the speech recognition system can be trained to understand the speech patterns of the translator, (ii) the signal-to-noise ratio is much better and (iii) the translator can skilfully "edit" his/her words to overcome any known problems. Such techniques are very promising in that they offer a cost-effective way of producing good quality live subtitling of difficult material, such as sports. However, one problem is that speech recognition software for the English languages is, at present, better than that available for other European languages. This disparity probably reflects the greater effort expended on software for the American market. Eventually, other languages will also benefit from the advances in speech recognition – but the gap might be 5-10 years.
From the perspective of deaf and hard-of hearing people, subtitling on analogue TV has some "technical" problems. Most video recorders cannot record subtitles. Subtitles also have to be switched off when "zapping" from channel to channel. Subtitles appear on different teletext pages (e.g. 777 in some countries, 888 in others). Fortunately, such problems can be solved by digital TV – where subtitles can be embedded in the picture (and hence easily recorded) and can also be permanently selected. Furthermore, page numbers are not used on digital TV services.
Even so, subtitling is not necessarily simple in the digital world. The DVB Project has developed two systems of subtitling: Teletext via DVB (ETSI EN 300 742) and DVB subtitling (ETSI EN 300 743). In the former, the DVB receiver reinserts the teletext lines in the vertical blanking interval of the analogue TV signal – thus ensuring that the subtitles can be decoded by almost any TV. On the other hand, DVB subtitling offers many benefits, such as bit-maps allowing complex fonts and text, more flexibility, more colours, better graphics, etc. Although DVB subtitling could be described as the "preferred technical solution", the danger is that when you have two standards for the same function, the market gets confused! In the UK, almost all DVB-T receiving equipment supports DVB subtitling. In other countries, many DVB set-top boxes simply use the "teletext via DVB" option and do not support DVB subtitling. Consequently, broadcasters implementing DVB subtitling in some countries have been disappointed to discover that the public cannot receive such services!
Audio description for blind viewers requires a separate audio channel carrying "narration". Several techniques have been developed to deliver audio description via analogue TV. In the mid-1990s, a European collaborative project AUDETEL developed a system which converted speech to data for transmission via teletext on analogue TV services. Although the resulting audio quality was marginal, the consumer electronics industry was concerned about the high costs of licensing some of the inherent technology and were reluctant to produce receiving equipment. Consequently, AUDETEL was abandoned before entering service. On the other hand, some broadcasters occasionally devote one channel of their "stereo" sound service to audio description, but this is not a satisfactory solution because viewers now expect stereo sound.
As digital TV can support multiple audio channels, each TV service could have the normal stereo sound at 192 kbit/s plus another stereo channel (including audio description) at 192 kbit/s – corresponding to a total bitrate of 384 kbit/s. The viewer selects whether to listen to the normal stereo sound or to the pre-mixed sound + audio description. This method works well for satellite services, but broadcasters using DVB-T generally do not have enough capacity to transmit all of the required TV services – adding more audio channels makes this problem even worse. Some viewers want the flexibility of adjusting the level of the audio description relative to the main audio channel. In practice, this requirement offers the opportunity to reduce the transmitted bitrate since the audio description can be coded at, say, 64 kbit/s – the resulting total bitrate would thus be 64 + 192 = 256 kbit/s. Of course, there is a price to be paid for this reduction in bandwidth: those wishing to receive the audio description service must have a special receiver that can simultaneously decode separate audio channels, together with the facility to mix the two signals. Adding such features to a DVB receiver chipset would add a trivial amount of complexity, but would bring great benefits to blind and visually-impaired people.
Sign language is the primary means of communication for some deaf people. Recognizing this need, some EBU Members offer "in-vision" video images of a "signer" during certain programmes, typically those of special interest to deaf viewers. Widespread use of this "in-vision" technique is limited by the fact that most viewers (i.e. those with no hearing impairments) are distracted or irritated by the signer, even if the insert occupies only about one-sixth of the screen area. It is clear that a workable solution must allow viewers to select whether or not they wish to see this supplementary service – as they can already do with teletext subtitles.
Unfortunately, the MPEG-2 video compression system currently used for digital TV broadcasts is not very efficient at coding small images. The result is that transmitting the separate video stream containing images of the signer can demand 2 Mbit/s or even more. More advanced compression schemes (such as MPEG-4 Part 10) can operate well at much lower bitrates, but this would require all users of the signing service to have special receivers. As the processing power required to decode the new compression schemes is significantly greater than needed to decode MPEG-2 services, the cost of such receivers is likely to be a problem for some years. One alternative might be to harness the power of computers which can easily be used to receive digital TV services and provide video overlays for signing.
As mentioned earlier, subtitling on analogue TV was established on the success of teletext. Hearing-impaired viewers did not have to buy TV sets specially designed for minorities. Audio description and signing may not have the benefit of being able to have a free ride on technologies used for other broadcast services. It is important to recognize that the economies of mass production will not be achieved if different countries pursue different technical solutions for audio description or signing. Technical standards need to be developed at the European level to ensure that blind and deaf people will be able to afford the necessary equipment.
If a member of your family was blind, how much would you pay to buy the hardware that would allow them to listen to audio description? My guess is that the costs of appropriate receiving equipment may not be the major problem. The real problem is that, at present, the hardware is not available at any price – and neither are the services! In the absence of agreed technical standards, it is understandable that manufacturers have not produced receiving equipment ... and broadcasters have not introduced such services.
Broadcasters are naturally concerned about the costs of providing these additional services. On the one hand, it can be argued that the costs of providing subtitles is a tiny fraction of the costs of making new TV programmes. How can a broadcaster complain that subtitles are too expensive at, let us assume, €600 per hour? If the subtitles were used by 30,000 people, the cost per viewer would be only €0.02 per hour.
On the other hand, providing 10 hours of subtitling every day would cost more than €2 million each year. Such amounts may be affordable for the large and successful broadcasters, but the reality is that many broadcasters (public and commercial) around Europe are desperately short of money to spend on programmes of all types. In very large countries, subtitling can benefit millions of people, thus the expenditure can be justified in terms of cost per user. In smaller countries, the cost per user is much higher. It is easy to see why many commercial broadcasters are reluctant to commit themselves to widespread subtitling.
EBU Members have long been active in technical developments supporting the needs of disabled people: subtitling, audio description and signing have all been initiated by EBU Members. Many EBU Members have already made substantial voluntary commitments to offer widespread services – for example, the BBC plans to subtitle 100% of its TV programmes by 2008. Few EBU Members have the resources to match the BBC's efforts, but all public service broadcasters have obligations to serve the entire community – which includes disabled people.
Now is the time for EBU Members to start planning how they will meet the challenges of serving the needs of disabled people.