This piece came from a weekly archive of NY Times and Washington Post
technology articles compiled by Will Smith -
Learn more or subscribe with a message to [log in to unmask]


-- forwarded article --

Now Hear This, Quickly

   By DOUGLAS HEINGARTNER

   "WE call it the 66-second minute," Laura Gaines said.

   Ms. Gaines is the vice president of Prime Image, a maker of devices
   like the Digital Time Machine that shorten audio and video recordings
   by up to 12 percent with "no discernible results." Micro-editing, as
   the process is called, created a stir last year when some broadcasters
   were reported to be using the technology to squeeze more
   advertisements into the same block of time.

   As it turns out, it was hardly an isolated phenomenon. Creating more
   time is the impetus behind many new technologies that allow listeners
   to pick up the pace.

   From call centers and intelligence agencies to radio stations and
   universities, such technology helps listeners try to keep up with the
   growing number of audio recordings piling up on the air, on the phone
   and on the Web. Wading though this mountain of words faster than it
   takes to say them not only saves companies money; it might help people
   absorb more knowledge.

   The new software programs, DVD players and phone services rising to
   this challenge all take advantage of the human ability to comprehend
   speech much more quickly than the typical spoken rate of 140 to 180
   words a minute. How many times as fast? "I've heard of instances where
   people go to 4X, and they still want it to go faster," said Blake
   Erickson of Telex Communications, which makes "talking book" audio
   players for the educational market.

   Scientists have long known that people can understand speech at a rate
   of up to 400 words a minute and beyond. "Speech rate isn't limited by
   the listener," said Arthur Wingfield, a psychology professor at
   Brandeis University. "It's limited by the speaker."

   In normal conversation, only a small part of the brain is taxed,
   leaving excess processing power to be used for listening for lurking
   predators, filtering out background noise or simply daydreaming.

   But speeding up speech on analog equipment like cassette decks
   traditionally led to the dreaded chipmunk effect, making long-term
   listening untenable. Digital time compression, however, works by
   discarding tiny segments of repetitive audio (for example, 30
   milliseconds of a vowel) and reconnecting the remaining bits, leaving
   the pitch unaltered.

   Simple versions of digital time compression have been available for
   years in devices like answering machines and hand-held recorders but
   did not offer much in terms of user control. A confluence of smart
   software, wider Internet access and inexpensive hardware, however, now
   enables listeners to choose when to step on the gas.

   Auxiliary programs, or plug-ins, that allow digital audio and video
   recordings to be played faster (or slowed down) at will have recently
   become available for popular software like RealOne and Windows Media
   Player. Perhaps the most popular is Enounce's 2XAV plug-in (which
   works with both Real and Windows players and costs $29.95); the latest
   version of Windows Media Player offers a proprietary version of this
   feature. Similar capabilities are finding their way into other
   hardware - for example, the latest DVD recorders from Panasonic.

   "You can watch a two-hour movie on a one-hour flight," said Chris
   Binace, an Enounce software developer. Yet this kind of software is
   not generally intended for entertainment listening. So far most
   end-user applications have involved academia, for example, allowing
   students to listen to archived audio or video lectures.

   Online, the amount of recorded audio is growing at an overwhelming
   rate, providing a new impetus for speed listening. A spokeswoman for
   National Public Radio said that demand for NPR audio on the Web was
   about 50 percent greater in June than it was a year earlier, and now
   averaged 5.5 to 7 million audio downloads a month.

   "You just have oodles of data,'' said Ed Rucinski, a vice president of
   the Dictaphone Corporation, "and if you can only listen to it in a
   real-time fashion, that's your bottleneck." Mr. Rucinski's company
   records "literally millions of hours" of audio every year: medical
   dictation, emergency calls to 911 centers, even financial
   transactions. "Any time you call your broker," he said, "that gets
   recorded."

   One company addressing the deluge is Fast-Talk Communications, which
   makes software for large businesses that scours voice and audio data
   much the way search engines sift through text. Many Fast-Talk clients
   work in intelligence. "But there's a limited number of linguists,"
   said Bob Crochetiere, a Fast-Talk sales engineer, so companies have to
   find ways of processing this material more efficiently. Mr.
   Crochetiere said clients would often listen to audio at speeds
   increased by as much as 50 percent, but only in bursts because after
   too much fast listening, "they start zoning."

   Hannah Hawkins, transcription manager for CCBN, a company that records
   and archives hundreds of lengthy conference calls each week for the
   financial industry, said, that speed was crucial. Clients need the
   transcripts as soon as possible after the call is finished, so CCBN
   transcribers sometimes double the playback speed of familiar portions
   like introductory legal disclaimers.

   "If they're speaking very slowly," Ms. Hawkins said, "you can
   understand them perfectly" at accelerated speeds.

   Richard Brownrigg, a general manager at [1]RealNetworks, which makes
   the RealOne media player, said that fast playback was still in its
   early days,but that he could imagine its value expanding as voice
   technology crossed into new areas. Playing back long cellphone
   messages in half the time, for example, becomes attractive "when
   people don't want to chew up their minutes," Mr. Brownrigg said.

   In advertising, where costly post-production of commercials can take
   longer than the production itself, the potential savings are vast. "To
   edit a 30-second spot can take half a day," said Ms. Gaines of Prime
   Image, but takes just minutes with the company's technology. (She
   hastened to point out that the compression was intended to enable
   advertisers to say more in the same period of time, not to let
   broadcasters shortchange the advertisers.)

   Most research has indicated no loss of comprehension or
   intelligibility at playback speeds of two or even three times normal
   speed. Cameron Earle, who is helping to commercialize variable-speed
   playback applications developed by Brigham Young University, said that
   most students chose rates that were 80 to 120 percent faster than
   normal with no decrease in test scores. Although it does take some
   getting used to, Mr. Earle said, he estimates that "80 percent of
   acclimation is in the first hour."

   Perhaps even more significant, the technology may have benefits beyond
   saving time and money. "People who are listening at accelerated speeds
   learn just as much, and there's some evidence they may learn even a
   bit more," said Kevin Harrigan, an associate professor at the Center
   for Learning and Teaching Through Technology of the University of
   Waterloo in Canada. The consensus is that the extra brainpower needed
   to follow speedy speech enhances comprehension. "If you're listening
   at accelerated speeds," said Joel Galbraith, a researcher in Penn
   State's instructional systems program, "it forces you to not do
   anything else, so you're more focused on it."

   Ray Juang, a University of California undergraduate who would often
   fall asleep in Berkeley's vast lecture halls, agrees. "On average, I
   understand the material better during playback than in the actual
   lecture room," Mr. Juang said. "The speed-up does force me to pay more
   attention."

   Accelerated speech also piques interest. A quarter-century ago,
   Priscilla La Barbara, a marketing professor at New York University,
   found that time-compressed radio advertisements were perceived as more
   interesting and led to higher rates of recall.

   But the days of those fast-talking radio announcers ("3.7 percent
   A.P.R.," "void where prohibited") may be numbered: Esther Janse, a
   post-doctoral researcher at the University of Utrecht, has found that
   digitally accelerated speech is more intelligible than the natural
   speech of a person talking rapidly. "When you try to speak faster and
   faster, speech gets very blurred," Ms. Janse said. The distinctions
   fade, she said, whereas digitally accelerated speech uniformly
   preserves all the crucial intonations and inflections.

   There are other examples of how machine-altered speech may trump that
   of humans. Professor Wingfield of Brandeis said that airplane pilots
   had been shown to pay greater heed to warnings issued by computerized
   voices than natural human recordings. "When one of these hokey
   synthesized computer voices says to pull up," he said, "it's like,
   'Oh, well, that's a computer. It must know better than I do.' "

   Synthesized accelerated has many other devotees. "When I listen to the
   newspaper, I tend to go as high as 650" words per minute, said Gregory
   Rosmaita, a Web designer based in Jersey City. Because Mr. Rosmaita is
   blind, his interface with computers is audio-based, in the form of a
   synthesized voice that reads text aloud. He prefers British English to
   American in this regard. "With the more clipped British speech," he
   said, "I can increase the rate even faster."

   He said he had become so accustomed to accelerated speech that normal
   rates could sound unnatural. "It's actually difficult to comprehend
   the speech when it becomes that slow," he said. "It's sort of like
   watching a marquee scrolling one letter at a time rather than one word
   at a time."

   Some users compared it to going back to dial-up Internet access after
   experiencing broadband. "I cannot stand to listen at 1.0," said Mr.
   Earle of Brigham Young. Mr. Galbraith of Penn State agrees. "Once you
   go faster, you just can't go back to real time," he said.

   There are some caveats: for example, the capacity to understand fast
   speech seems to fade with age. "The younger the person is, the faster
   they can go," said Mr. Earle, who said he had noticed a drop-off
   around age 30. "Professors can never go as fast as the students.
   Students can crank it out."

   Few question that rapid playback saves time. "There's no doubt,
   absolutely," said Patrick McClanahan, a Navy lieutenant commander who
   used variable-speed playback while earning his master's degree in
   business administration at the Wharton School. Commander McClanahan
   said he most appreciated the ability to find a crucial point in a
   recorded lecture. "It's virtually impossible to slide that little
   thing across and find exactly what you want," he said of the cursor in
   audio playback software. Variable-speed playback eliminates the need
   to do so.

   Mr. Juang, who as a Berkeley undergraduate has sometimes watched six
   two-hour lectures a day, said that even with occasional buffering
   delays and the need to replay bits that went by too fast, "an hour
   takes 35 or 40 minutes at most."

   So as fast listening becomes commonplace, will more people turn into
   fast talkers?

   "We're used to hearing things faster, so it probably translates into
   our talking as well," Mr. Galbraith said. "We'll start conditioning
   ourselves to just expecting and needing it faster."

   Professor Wingfield of Brandeis is not so sure. "Knights were jousting
   with the same brain that we're using today," he said. "The
   articulatory system, the physiology of speech has not changed."

Posted by
Mark Senk | 412-386-6513 | [log in to unmask]


VICUG-L is the Visually Impaired Computer User Group List.
To join or leave the list, send a message to
[log in to unmask]  In the body of the message, simply type
"subscribe vicug-l" or "unsubscribe vicug-l" without the quotations.
 VICUG-L is archived on the World Wide Web at
http://maelstrom.stjohns.edu/archives/vicug-l.html