Measuring the quality of live subtitling: Statement
1.1 Earlier this year, Ofcom consulted on proposals to improve the quality of live subtitling on UK TV to benefit deaf and hard-of-hearing viewers, among others.
1.2 Ofcom would like to thank the subtitle users, broadcasters, subtitle providers, representative bodies for hearing-impaired people, and others who responded. We have carefully considered their comments, which are summarised in section 2, together with Ofcom's responses to them, including our decisions. These decisions are summarised below.
Measurement of quality
1.3 In the light of responses to the consultation, and of the information available to Ofcom, we have decided that broadcasters should be required to measure the following dimensions of quality, on the basis of samples of live subtitling selected by Ofcom:
a) the average speed of the subtitling;
b) the average latency of the subtitling (the delay between speech and live subtitling), and the range of latencies; and
c) the number and type of errors (i.e. minor spelling errors, major omissions or factually misleading subtitles).
1.4 For this purpose, Ofcom will identify samples of live subtitling in three genres of programming - news, entertainment and chat shows - and ask broadcasters to carry out measurements. We shall ask broadcasters to collect data using a model which has been used for some years by academics, broadcasters and subtitle providers to categorise errors. Preparation for this will assist in the measurement of speed and latency. We will arrange a briefing for those involved in preparing for and carrying out the measurements. We intend that the measurement exercise should be carried out at six month intervals, starting later this year, for a period of two years.
1.5 We agree with respondents who said that ensuring consistency of measurements will be important. With this in mind, we approached the University of Roehampton, which has conducted research on various aspects of subtitling quality, and has applied the model in this context. Following discussion, Ofcom and the University of Roehampton have agreed that a small team from the university will validate the measurements provided by broadcasters from an expert, third party standpoint.
1.6 Once the validation exercise has been completed, Ofcom will clarify any outstanding points with broadcasters, and produce the final report. Ofcom intends that, if possible, the first round of measurement should be completed in time for the outcome to be included in the access services report due for publication in spring 2014.
Review of guidance, and possible targets
1.7 It is clear that there would be merit in looking again at the guidance in the light of the research conducted in recent years, but Ofcom considers that it would be better to await data from the measurement exercise. Ofcom will then consider what changes to the current guidance on latency and speed may be appropriate, and whether targets should be set. In doing so, we will also have regard to relevant research, and the views of interested parties.
Short delays in transmission to improve quality
1.8 Ofcom invited consultees to tell us about the factors that might facilitate or hinder the insertion of a delay in live transmissions sufficient to improve the quality of subtitling. Responses suggest that a delay of 30 seconds or less could allow live subtitles to be synchronised (resolving the problems of latency), corrected for errors (dealing with the problems of inaccuracy), and presented in block rather than scrolling format (making the subtitles easier and less time-consuming to read). A shorter delay would allow subtitles to be displayed in block form.
1.9 Broadcasters were strongly opposed to inserting delays, citing the risk to viewer trust, the need for complex technical solutions, and competition from other media, amongst other reasons. We note that some broadcasters consider short delays editorially justified to safeguard viewers from the possibility that they might hear obscenities. We invite broadcasters to consider whether, in some programmes which are not time-sensitive, delays could be justified on editorial grounds to enable the quality of subtitles to be improved significantly and look forward to maintaining an ongoing dialogue with broadcasters over this issue.
Late delivered programmes
1.10 Ofcom's consultation drew attention to the problems that occurred when pre-recorded programmes were delivered to broadcasters too late to allow subtitling to be prepared in advance, necessitating lower quality live subtitling.
1.11 We have asked broadcasters to let us know in January next year which late-delivered programmes had to be subtitled live between July and December 2013. We plan to include a report on this in the first report from the quality measurement exercise, due in spring 2014.
Block vs scrolling subtitles
1.12 Ofcom sought views on whether block subtitles or scrolling subtitles were better for viewers. There was a clear consensus that block subtitles are easier for viewers to read and allow them to spend more time looking at images. However, it is equally clear that viewers do not want this to come at the expense of increased latency.
1.13 It has been suggested that, with training, the process of preparing and displaying block subtitles may take only marginally longer than transmitting scrolling subtitles. In the light of this, Ofcom suggests that there would be merit in broadcasters experimenting with the use of block subtitles for some programmes (e.g. late night programmes on rolling news channels) and testing these with focus groups to see their reaction. We look forward to maintaining an ongoing dialogue with broadcasters over this issue.
1.14 In the meantime, Ofcom suggests that broadcasters should consider:
a) using block subtitles whenever possible; and
b) reversioning repeats of programmes with block subtitles synchronised with the original speech, in place of the original scrolling subtitles.
We look forward to maintaining an ongoing dialogue with broadcasters over this issue.