DTV brings better pictures, better sound and more channels and features. But closed captioning, the stalwart scroll of text that accompanies dialog, has proven trickier than many broadcasters—and their deaf and hard-of-hearing viewers—might have once figured.
Closed captioning has been around for more than 20 years, and has been required on most broadcasts and cable presentations since Jan. 1, 2006.
“I thought, like most people, that the whole closed captioning thing was wrapped up,” said Carl Dempsey, president and CEO of encoder/decoder provider Wohler Inc. in Hayward, Calif. “I didn’t think there was a market whatsoever for new captioning products.”
There are plenty of problems, it turns out. June 27, the FCC Disability Access Working Group reported numerous flaws and foulups in a report to the FCC Consumer Advisory Committee. The group and advocates for the deaf and hard-of-hearing have pointed to muddled, misspelled, mis-sized, misplaced and poorly timed captions—particularly in live programming such as news and sports, but also in some taped programs.
On the other side, scores of mostly small and independent programmers, ranging from churches to producers of fishing shows, have sought exemptions from FCC rules (promulgated under the Americans With Disabilities Act), citing the financial burden that compliance would cause.
WHERE’S THE PROBLEM?
| Poorly placed and misspelled captions are among consumers’ complaints|
With the increased complexity of DTV facilities, and with the many parties (networks, affiliates, cable operators) involved in delivering a single A/V feed to viewers, there are many places—from the original program to the home television or set-top box—where bugs can arise. Consumers have complained that the various players point the finger at one another.
Captioning is moving toward the EIA-708 standard for the digital era, which involves more features, but also requires more bandwidth and creates more opportunities for error.
Most live captioning is still done by humans in the style of court stenographers; they hear the speech, mostly by telephone, and type the captions for encoding into the main broadcast signal. Voice-recognition technology has not yet achieved much acceptance in closed captioning.
So, many equipment makers are improving their captioning products and services.
Wohler has adopted a “service-oriented” approach with its HDCC-220A closed captioning bridge, working closely with customers to customize the two-channel encoder/decoders to each particular facility. “We say, even if it’s just a block diagram, tell us what you want to achieve,” said Dempsey. “We don’t even send a demo without getting some information from the client. Our mantra is, we want to be an extension of our customer’s engineering department.”
EEG, a captioning pioneer, noted that most broadcast monitoring systems don’t adequately monitor captioning; just watching what leaves the station may or may not give an accurate picture of what’s appearing on viewers’ screens. So the company focused on some 35 common caption anomalies in creating its CB512 DTV caption legalizer, that fixes numerous problems and accurately monitors and creates a record of what really happened. In addition, stenographers have faced the problem of low audio quality on phone lines, or latency using the Internet, so EEG developed a downloadable client in its new HD480 captioning product that allows captioners to actually see what’s coming out of the encoder, instead of captioning straight to air.
Also available for the HD480 is a module to write audio metadata in the HD VANC (vertical ancillary) space, simplifying routing and control.
Computer Prompting and Captioning (CPC) is touting its software-based captioning system for less that $6,000 that works on both Macs and PCs—a solution for any programmer with 10 or more hours to caption. It allows operators to type the caption using a foot pedal to start and stop the audio. The company also makes hardware-based solutions and software that plugs into nonlinear editors.
XOrbit offers hardware systems for those with lots of captioning, and CaptionSmart, an Internet-based service for programmers, post houses and anyone else with a smaller load of programs or spots to caption. The user sends its MPEG-2 file (SD or HDTV) online, along with the text file of the caption, and the service sends it all back. CaptionSmart can turn around a 30-second spot in just a few minutes, or a 30-minute program in a couple of hours—all for about $3 per minute of program time. For stenographic services, XOrbit offers RTX, which delivers compressed versions of the on-air video so captioners can see as well as hear the speakers, which enables not only greater accuracy but lets them place captions in the correct spot on the screen.
Voice-recognition technology for live captioning is not yet widespread. But Michigan-based Enco has developed EnCaption, a system that relies heavily on pre-programmed context (local place names, politicians and sports heroes, for example) to improve accuracy. But product acceptance has been slow, and the need for context explicitly creates delay in the captioning.
Comprompter is a fourth-quarter release of Caption Central, which works with its NewsKing automation system and all others tested so far, the company says. It fuses live voice recognition with pre-scripted text to let news reports include ad libs and other unscripted content in the closed captioning. It includes an audio mixer that allows live studio voices to speak through heir own channels directly into the system and produce captioned text. It also spell-checks with Microsoft Word. Comprompter CEO Ralph King said the voice system works more quickly than typing the words and can produce accuracy as high as 95-96 percent live.
Several estimates say 10 percent of Americans—some 30 million people—are deaf and hard-of-hearing.
(Part Two of this article will address the problems in captioning encountered in HDTV broadcasts.)