This report examines the issues involved in making video-on-demand (VOD) programming accessible to viewers with disabilities. For this report, audiences of interest include deaf and hard-of-hearing people, the blind and visually-impaired, and people with certain mobility impairments.
Though this report was produced... for the Canadian Broadcasting Corporation and based on the CBC’s own VOD offerings, the report’s findings may be useful to broadcasters worldwide who have an interest in opening up video-on-demand programming to this audience of people with disabilities.
Making VOD accessible to viewers with disabilities is a win-win proposition by:
Accessibility on broadcast, cable, and satellite TV is well established, but all-digital services like VOD are new enough that not every problem has been solved. While captioning is straightforward, audio description and onscreen menu systems remain a barrier because digital delivery systems were not designed with accessibility in mind. Upgraded versions, complete with accessibility features, may eventually be introduced, but the industry needs working systems that use the current technology to demonstrate what can be done today and what needs to be improved tomorrow.
Even though the digital delivery infrastructure is comparatively new, when it comes to accessibility, those systems amount to a kind of legacy platform that requires exploration and innovation. Adding accessibility features to VOD is a way to show leadership: “We made it work first.”
This report provides two possible models that work around the current inability to make onscreen menu systems truly accessible to blind and visually-impaired viewers. Statistical facts, workflow options, and cost quotes are also provided.
The conclusion? It is practicable to provide accessible programming on VOD, and doing so provides a host of benefits to the broadcaster, the viewer, and the industry at large.
First, how large is the audience of people with disabilities?
Statistics Canada’s figures, published in 1994, estimate the number of people with two disabilities related to television viewing. The numbers are old (they’re based on the 1991 census) and won’t be updated until at least December 2003. StatsCan’s definition of mobility impairment is unrelated to VOD usage, so those figures are not cited here. Some figures are approximated in the original Statistics Canada documentation.
While people with disabilities are a minority, even within the city of Toronto the populations are not small.
What are the viewing habits and TV/VCR ownership patterns of people with disabilities?
The Who’s Watching? study by the American Foundation for the Blind found the following:
|Issue||Blind audience in survey||“General population”|
|Own a television||99%||99%|
|Own a VCR||83%||85%|
|Watch television at least two or three times per week||97%||95%|
|Mean number of hours of television watched per week||24||29|
No similar figures are available for Canada, but this report assumes Canadian results would be broadly comparable.
For deaf/hard-of-hearing viewers, a survey by the Canadian Hard of Hearing Association found:
- CHHA members watch more TV than the general Canadian public: 20.9 hours per week on average versus 13.8 hours
- CHHA members are more likely to be classified as “heavy TV viewers” (68% vs. 46%)
- Canadians with television viewing habits notably higher than the average (13.8 hours per week) tend to be:
- Those with a hearing loss (15.7 hours)
- Aged 55 years and older (18.4 hours)
Apart from the CHHA study mentioned above, there is little evidence on the use of cable TV, including digital cable, by people with disabilities. The Canadian Cable Television Association (PDF) reports 1,286,000 digital-cable customers in 2002 (17.6% of cable-TV customers). It’s possible that deaf and hard-of-hearing people have similar rates of uptake of digital cable, though usage is likely to be lower for blind and visually-impaired people due to the lower accessibility of TV for that group.
In the U.K., a study, if it can be called that, of ten subjects found that seven had digital cable.
The barriers faced by people with disabilities are easy to state:
What techniques are used to make TV accessible to viewers with disabilities? First, let’s define our terms.
Almost all captioning on television is closed, meaning you the viewer have to turn it on. In Canada and other countries using the NTSC television format, Line 21 of the vertical blanking interval carries caption codes. (That line is just above the visible part of the picture.) You need a decoder to turn the captions into visible words. Since mid-1993, decoder chips have been standard equipment on nearly all new televisions. At least 13 million decoder-equipped TV sets have been sold in Canada since then. In all likelihood, viewers with the digital set-top boxes needed to watch VOD programming have TV sets with caption decoders built in.
Any television signal, using any delivery mechanism or tape format, can be closed-captioned. For MPEG-compressed video, in some circumstances (e.g., DVD) the encoding station must be configured to save Line 21 into a special packet. In other encoding methods, Line 21 is automatically captured.
Audio description for blind and visually-impaired viewers is delivered as a separate audio track containing the entire main audio plus the description narrator. That audio track is usually recorded on Channel 3 of a tape. The main-audio-plus-description mix is delivered by second audio program (SAP) on analogue TV stations.
For digital services, whether at reception in the home (e.g., digital cable or direct-to-home satellite) or origination at the broadcaster (e.g., an all-digital specialty service), getting a described audio track to the viewer is not easy. There is no SAP in digital delivery. While there may be alternate audio tracks in certain systems, they aren’t coded as second audio program, meaning that described audio on SAP doesn’t automatically pass over to an alternate audio track.
In any event, service providers have not chosen to use alternate audio tracks for description. The two options currently favoured to deliver described audio over a digital service are:
It appears there is little for a program producer to do that will increase accessibility for people with mobility impairments (chiefly the use of hands and arms). The issue becomes the remote control and other physical switches used to operate the equipment. While certain kinds of adaptive technology could be used in those cases, there seems to be nothing CBC or its partners could do to improve accessibility for this group in, say, development of menu interfaces. It appears to be a question for hardware manufacturers.
Another barrier faced by blind and visually-impaired viewers is graphical menu interfaces. Every system in common use relies on a visual menu to activate and deactivate functions that are more advanced than turning power on or off or changing channels or volume. Some unusual devices, like a few TVs, have dedicated keys on remote controls for useful functions (like turning on captions or SAP), but even the labels on those keys are visual. If you’re blind or visually-impaired, either you find it hard to use these interfaces or you can’t use them at all. That’s one reason why no digital service in Canada uses alternate audio tracks to deliver description – you need to be able to see to turn the alternate audio tracks on.
Some guidelines have been published on creating accessible audiovisual menus (that is, with all the usual graphics and additional voice prompts), but they’re not very solid or easy to implement. Apart from a tiny handful of DVDs, there aren’t even any examples to point to; there is no known set-top box or other digital TV service with audiovisual menus.
In this section, we’ll examine specific issues relevant to VOD accessibility and document our findings for each.
If the goal is to provide more programming with captions, how easy is it to recover or capture the captions on material that was previously captioned and will now be rerun on a VOD service?
The assumption here is that programs will be repackaged or re-edited slightly for VOD; if the program airs on VOD in exactly the same form as on another network, then the same captioned submaster used for that other network can be reused. (An example would be a rebroadcast of The National on VOD. It’s simple to record the original with intact captions and digitize that tape for VOD.)
CBC reports it has had no recent luck whatsoever in recovering previous captions. It’s technically possible to recover the text of captions, assuming a closed-captioned submaster tape can be located. (You’d use a data-recovery decoder to download the caption text into a file on a computer.) But capturing the text doesn’t capture the position or timing codes; at best you get 1/3 the information you really need. And sometimes it’s impossible to find the captioned submaster.
Even sending a clone of a tape back to a captioning house to be recaptioned from the source file doesn’t always work. Captioners occasionally lose the original caption file or never preserved it in the first place.
Moreover, there is no such thing as a universal caption-interchange format. The Open & Closed Project aims to solve that problem eventually (see the PDF that documents the Accessibility Exchange Format™), but as it currently stands, each “authoring tool” (captioning software package) produces its own file format, most of which are incompatible with each other.
Assume that recovery of captions on previously-captioned material will be impossible for the foreseeable future. All programming aired on a VOD service will need to be captioned from scratch, with the exception of programming aired from a captioned tape that is unaltered from an original source.
For CBC Television and CBC Newsworld, which must caption the entire broadcast day save for outside commercials, captioning comes from four sources:
CBC’s in-house captioning department works at or near capacity to maintain captioning requirements for CBC Television and CBC Newsworld; they work mostly on quick-turnaround material like promos. Almost any quantity of new material that department would be asked to caption for VOD would crowd out material for the other two networks, forcing that material to be sent out-of-house and, more importantly, putting stress on the quick deadlines the in-house department must meet.
CBC’s captioners have captioned items for other CBC departments before, but only occasionally, and the cost of captioning is always billed internally (at $165 per program hour, prorated linearly).
CBC’s typical out-of-house suppliers use scrollup captioning for everything, even fictional narrative programming. While scrollup captions are generally agreed to work fine for most news and current-affairs programming, they are frowned upon for documentaries and fictional programming, where pop-on captions (individually-timed and -placed caption blocks) are more proper and enjoyable to read.
Assume VOD programming will be sent out of house for captioning at billable rates in the $165/hour range. Since VOD will ultimately be a premium service, CBC may opt to use different captioners and rely on pop-on captions as much as possible.
VOD programming that is finalized weeks in advance presents no scheduling problems for captioning. Weeks-long leadtimes border on the luxurious in captioning, and any competent captioner would welcome such jobs and easily meet deadlines.
Some VOD programming becomes available to viewers one or two days after it’s finalized and packaged. CBC’s in-house captioning department finds that turnaround time rather tight, even for 15- to 20-minute segments. But outside captioners should be able to meet those deadlines, especially given the smaller quantity of material. Captioners should be discouraged from charging a premium for the shorter-turnaround jobs given that such a turnaround isn’t really short in captioning terms – commercials and music videos often have to be turned around in three hours or shorter.
Little or no truly live programming is contemplated for VOD, so real-time captioning is not required. If it ever became necessary, though, the remote captioner would have to be able to watch a first screening of the programming to create the captions on a stenotype keypad. The resulting tape could be digitized for VOD, but it would be better to clean up the captions and digitize that set of captions. In practice, that would end up being nearly indistinguishable from conventional “offline” captioning (of prerecorded programming), making any use of real-time captioning for VOD even more unlikely.
Assume that the VOD service’s current deadlines for material can be readily met by captioners. Resist any effort to charge premium prices for shorter-turnaround jobs. Assume that real-time captioning will rarely, if ever, be necessary for the service.
There is no indication that encoding staff need to do anything special to digitize a captioned tape into MPEG format compared to an uncaptioned tape. Captions are automatically preserved.
CBC’s in-house captioners work in English only. It’s easy to find outside captioners who can caption French-language programming, but quality is even more of an issue there than in English. Nonetheless, French programming can be and is captioned every day. There is no intrinsic barrier preventing the addition of captions to French-language VOD programming, should any be offered.
The same guidelines applied to captioning for English-language VOD can be applied to French-language VOD.
The bad news: On the current VOD platform, there is no way to associate sounds with menus. There is no way to make a talking menu system using the platform’s own commands and protocols. Audiovisual menus are impossible at present. That’s the menu or interface side of the problem.
On the programming side, there’s the question of getting a described audio track to the viewer. There aren’t all that many options, but some of them may work.
The first question to decide is: One VOD channel or two?
In conventional broadcasting, a single feed must serve everyone. You’ve got one CBC Television and one Newsworld feed, for example, on your TV set. With a single signal that has to serve everybody, captions and descriptions must be hidden in the signal and turned on by the viewer.
In VOD, though, the same constraints do not hold. Programming without description can appear on one station and with description on a separate station. The viewer simply dials up channel 350 to watch without audio description or channel 365, for example, to watch with audio description.
Adding this new channel isn’t as resource-intensive as it would be in conventional broadcasting. Virtual channels are believed to be possible possible with the current system. At the back end, the cablecaster merely has to set up a different decision tree for the new described virtual channel that plays only the described tapes.
In this case, we could adjust our offerings to work around the inaccessibility of the menu system. We could offer one described show on the virtual channel. You would tune to that channel and it would simply run – no menu interface required. We could rotate the programming every day or week so that the viewer at home could simply come back to the station after the next rotation to watch the subsequent program. Since the service is video on demand, the viewer could call up the programming at any time of day that’s convenient within the broadcast window.
That wouldn’t be the same as user-selected access to a range of programming, but, over a period of time (say a broadcast month), the effect would be the same: The blind viewer could watch described shows.
The other option involves duplicating the model from conventional broadcasting and using a single channel to serve everyone. Here, ideally we’d provide a command to turn on audiovisual menus so that the blind or visually-impaired viewer would know what the options are and make choices accordingly. But as we know, there is no way in the current platform to add sounds to a menu.
To make a single feed work, we’d have to hack the interface somewhat. We could set up a key command that would trigger a different decision tree and automatically run a described program. Here, the blind viewer wouldn’t hear an audio interface; the key command would, without further discussion, start a described program rolling. In this model, only one described show at a time would be available, but they could be rotated every day or every week, for example. You’d also have to know the key sequence up front.
This kind of pragmatic accessibility isn’t ideal, but it’s significantly better than what we have now.
If we’re hacking the interface, though, we could go further. We could set up several keystrokes, each of which would run a separate program. The A, B, and C keys on the Rogers Digital remote are useful for this purpose. We could make three described programs available at a time.
But in this model, there is no way for the VOD service itself to tell viewers what programs are described; we don’t have talking menus at our disposal. We’d have to use different methods:
Neither the single-feed nor the virtual-channels option is genuinely better on its own merits. Both require technical (re)programming. The single-feed option is likely to be cheaper because it will not require the creation of a new virtual channel. Assume the single-feed option is better for that reason.
Adding descriptions is a more time-consuming task than captioning, but some service providers canvassed for this report stated they could meet even the short deadlines in the current VOD service.
An issue arises, however, with captioning. There is no recent practice among broadcasters in Canada, the U.S., or the U.K. to present a program with audio descriptions but without captions. The captions and descriptions serve mutually-exclusive audiences, but in conventional broadcasting, a single tape has to serve everyone (nondisabled, blind, deaf), meaning that a single tape must contain both accessibility features. The workflow for conventional broadcasters involves the following choices:
A case could be made that programming that a blind viewer explicitly opts into doesn’t have to be captioned; the audience there isn’t deaf. The workflow could run as follows:
In the early stages of delivering accessible programming, tapes with description but no captioning will suffice, but over the longer term, procedures should be developed to place captions and descriptions on the same submasters. That is the current industry practice, and doing so future-proofs the VOD service for later reuse of the same material. (You won’t have to go hunting for two separate tapes.)
A workflow in that case could look like:
|Receive captioned submaster with main audio||Tape 1 (CC, main audio)|
|Dub the captioned submaster||Tape 2 (CC, main audio)|
|Received described submaster||Tape 3 (no CC, described audio)|
|Re-stripe Tape 3’s described audio onto Tape 2||Tape 4 (CC, described audio)|
|Encode Tapes 1 and 4 onto VOD||Captioned VOD assets, one with main audio, another with described audio|
Assume that described programming doesn’t also have to be captioned at the outset of an accessible VOD service, but workflow should be developed to include both features on individual tapes.
The technical methods to deliver French-language description are the same as English-language methods. However, there is no known audio description on TV in French. (A few home videos have been described in French.) There might be two description providers in Canada at least nominally capable of describing in French. The concept is as yet untried, but there is nothing intrinsic to the French language that makes it impossible.
The concept is as yet untried, but assume there is nothing intrinsic to the French language that makes describing French programming impossible.
The case could be made that the programming currently offered on the CBC VOD service – news and current affairs – does not really require description. However, the available evidence on viewer preferences for description does, in fact, support adding description to all forms of programming.
The Who’s Watching? study by the American Foundation for the Blind found the following demand for description in two surveys of blind and visually-impaired viewers:
|Genre||Survey 1||Survey 2|
|Dramas or mysteries||85%||83%|
|Nature or science||67%||72%|
|News and information||61%||68%|
|Music programs or videos||40%||44%|
|Daytime soap operas||21%||28%|
No genre or category of programming should be exempted from description, but place the emphasis on programming that is more documentary in nature. Programming that straddles both “news and information” and “nature and science” should have highest priority.
Every known TV description service provider in the English-speaking world was asked to quote for describing tapes in the following categories, taken from an actual VOD schedule:
Responses were as follows:
|Describer A (Canada)||$850||$1,600||$3,100|
|Describer B (U.K.)||£220||£504||£945||£11/minute quick turnaround, £10.50/minute otherwise|
|Describer C (U.K.)||$434||$1,042||$1,953||$21.70/minute|
|Describer D (U.S.)||US$1,498||$2,995||$4,493||US$2,995/hour in quarter-hour increments|
|Describer E (U.S.)||US$950||$1,900||$2,850||US$1,900/hour|
Other providers either didn’t respond (including two in Canada) or gave price quotes that were unworkably vague.
Audio description is more expensive than captioning. Provide at least some limited described programming.
Partners who could assist in solving the problems at CBC and industry-wide include national and regional disability organizations. The best course of action is to start a pilot project with a small number of subjects to gain field experience, as the BBC did with its usability study (PDF).
An option here would be a joint venture among CBC, other broadcasters providing similar VOD services, and the Open & Closed Project to start a test experiment with real programming and real viewers with disabilities. We would then have a working prototype service to examine (and to show to the broader industry) and actual viewers whose experiences could teach us what works and what doesn’t, and what we need for the future.
Accessible VOD services should be aggressively promoted in the news media. Comparable VOD services aren’t doing anything innovative in accessibility; CBC could stand out from private-sector competitors by highlighting its technical achievement and its plan to meet the needs of an under-served audience.
It’s technically possible to caption 100% of current CBC VOD programming, and captioning is not very expensive. It’s also possible to describe 100% of CBC VOD programming, but not all of such programming urgently needs description, and costs are notably higher.
Provide high levels of captioning of VOD programming (90% seems attainable in the short term) and at least some described programming.