Voice
Communication in Business Volume 2
Essays on telecommunications,
1981-2002
The telephone has two
major problems: sometimes people don't answer, and sometimes they
are already busy on another call. Needless to say, great minds, with
new technology suddenly available to them, sprang upon these
problems. I wrote this for Ian & Lis Angus's
Voice Data Report of
October, 1985.
Thoughts On Voice Mail
(Voice-Data Report, 1985)
Voice mail has been developing rapidly, both
in residential and business service. In its simplest form, it uses
nothing more elaborate than a telephone answering machine; when the
person you want to reach is not touchable, the answering machine
takes your message for later delivery. It is interesting to note
that we have now come full circle: whereas we used to complain that
we simply could not talk to a machine, we now become annoyed upon
hearing a ring-no-answer when it is necessary to get a message
through. Some even consider it rude of others NOT to have an
answering machine.
Of course, you have to have the right kind of
answering machine; if it cuts off the caller after 20 seconds,
permitting only a call-back message, it is no better than reaching a
message center or answering service. Because it is impossible to
leave a lengthy message with a human being (if you doubt me, just
try it), reaching a recorder which will accept anything you say and
get it right is a great relief. The better machines are voice
controlled--as long as you keep talking, they keep recording. Others
permit a message of two to five minutes, maximum. In one minute,
talking at a normal rate you can transmit between 150 and 200 words;
roughly the equivalent of one typed page, double spaced, greater
than the length of most business letters. This is usually sufficient
to leave even involved messages, and once you get used to it, you
may be surprised to find how little information actually requires
two-way interaction. With two minutes of long distance time worth
something less than a buck and a half, it sure beats the overnight
services, both electronic and paper.
But the big thrust in Voice Mail today,
particularly in business, is centralized voice storage systems.
Based on the present very low cost of hard disk memory, fabulously
expensive equipment converts voice to a digital format for storage.
Why would someone spend $60,000 or more for such a machine which
uses just about the most inefficient method known to science for
storing speech, when regular answering machines of quite good
quality cost somewhere between $100 and $200 each? (At 16 Kb/s, our
one minute message takes up almost a million storage bits, while 200
words stored in ASCII characters need only 10,000 bits).
For one thing, a telephone answering machine
is great for ring-no-answers, but it cannot handle calls that come
in while your line is already busy on another call. For people with
only one line, call waiting solves the "busy" problem, but the call
in progress is, of necessity, interrupted. Where call forwarding on
both busy and no answer is available, as in a modern PBX, a shared
system can serve a number of people better than any number of
answering machines could. (At $200 a copy, answering machines break
even with a small centralized system at somewhere around 300
mailboxes).
But once you have a centralized system, it
can do even more. Although digital voice storage is inefficient, it
lets spoken messages be treated like any other computer file. You
can find one message among many, for one of many users, using
relatively simple search techniques. And you can use computer
addressing techniques to re-route that message to one or several
people, with or without appended comments of your own. Or, you can
use the system like telephone dictation to deliver a message to an
individual or a group.
There is one major problem with centralized
systems, however. You cannot look at them, as you can at an
answering machine, and see that you have a message. If the PBX or
Centrex switch can light a message waiting lamp on the called
party's phone to create awareness of a stored message, then we are
almost back to where we were with an answering machine. It is well
known that no message system is very effective when people are
unaware that it has messages for them. You cannot expect people to
call in to check; some will and some will not. But every time
somebody calls in and finds no messages waiting, "negative
reinforcement" will slow future call-ins and delay the timeliness of
message delivery when something is actually present.
In any event, a centralized message system
requires the active cooperation of the switch serving the user if it
is to be effective. It must be able to tell the PBX or CO switch to
light the user's message-waiting lamp, and it must encourage the
switch to tell it the mailbox in which to store an incoming message.
For the latter, "off-net call forwarding" can be used to call the
storage system via a tie-trunk.
A PBX with this feature needs no further
software to simply seize a tie-trunk to the voice storage system and
outpulse the digits that it thinks identify a distant telephone. The
voice storage system, of course, handles the incoming digits the way
a PBX handles DID, but uses them to identify a mailbox. (Note that
ordinary call forwarding, to another extension on the PBX or Centrex
which might be used by the voice storage system, is not sufficient
to identify to the extension originally called).
For a complete stand-alone voice mail system,
message-waiting lamps can be omitted by allowing the mail box call
back periodically. It can then deliver the message when the callback
produces an answer (supervision for which can be returned on a
tie-trunk, if not a CO trunk or extension). There is no law, of
course, that says the mail-box will be able to tell the janitor from
the intended recipient.
From the above, it can be seen that a
stand-alone voice storage system is possible when it deals with
messages and commands from the user directly, and one that uses
relatively standard PBX features to enable switch-mailbox
communication is somewhat better. But the best solution is to have a
data link between the PBX or Centrex control and the control of the
voice storage system. Then extensions rather than tie-trunks can be
used, and much better integration can take place. Here, of course,
is where ISDN with its CCITT No. 7 signaling would be ideal. The
alternatives are too much like 45 vs LP, VCR vs Beta, and laser vs.
stylus on video disks to want to even think about.
Carried to the logical extreme, disk-based
message storage systems can send messages to each other. With voice
reduced to nothing more than digital data files, one machine can
call another up in the middle of the night when the rates are low or
the tie-trunks are unused and apply standard data transmission
techniques to send information stored on one machine to mailboxes
located on another. Voice store-and-forward.
These systems have a lot to recommend them,
and voice synthesized prompts and other assists from modern
technology make it possible for a stranger to wander into to an
answering-machine situation and still respond properly.
Unfortunately, many of the features designed for use by "insiders"
derive from the mentality that gave us the "flash and feature code"
approach to "modern" features on single line PBX phones. So far, I
have been unable to figure out all the tricks that have to be done
with *, # and various one, two and three digit codes to get the
thing to back up and replay, let me insert a few words, make a
message private, etc.
Voice mail systems have been discussed and
developed for a number of years. The irrepressible Ray Kraus, of
Consulting Communications Engineers, Inc., made a rather elaborate
proposal for "Voicegram Serivce" back in 1971, and talked about it
at many IEEE meetings and in other technical forums. Ray felt that
the system had to be centralized and run as a public utility, a not
unusual conclusion after his 30 or more years with Pennsylvania
Bell. However, Bell Labs seems to have agreed with him; they
developed their Voice Storage System, or VSS, and had it up and
running in a Philadelphia central office in 1980. It worked fine
with residential and Centrex lines, but it couldn't deal with PBXs;
ATBs are rare on CO trunk groups, and, except on DID trunks, you
have no way of knowing which extension a message might be intended
for.
The VSS is no longer in operation; the
Telephone Answering Industry recognized its threat to their
whip-socket business and went to court. But AT&T is now offering
Audix in connection with its PBXs.
One of the big guns in voice mail systems is
VMX of Richardson, Texas, run by Gordon Matthews of WATS-Box fame.
VMX and Wang both announced disk-based voice-mail systems in the
summer of 1981, and IBM followed a few months later, but VMX seems
to have done quite well both as a stand-alone and as a system
available on an OEM basis to many PBX manufacturers. The VMX (Voice
Mail eXchange) is a well thought out machine with several years of
real live experience in the marketplace; some PBX designers simply
do not have time to reinvent the wheel, and welcome the chance to
incorporate a highly desirable subsystem into their equipment. But
there is another reason why VMX is doing well: Patents. VMX holds
U.S. Patent 4,371,752, issued in February, 1983.
The July, 1985, issue of High Technology
discusses the VMX patent situation at some length and, indeed,
regular readers of Electronic News and other industry
journals know that VMX has gone after many smaller companies who are
trying to get into the voice mail business. "Pay our license fee or
go to court" seems to be the general idea, and the smaller companies
simply cannot afford litigation. A recent VMX press release
identifies Opcom as the 13th company to agree to pay royalties for
using the VMX patent. VMX hasn't gone after AT&T, but an agreement
has been worked out with IBM to cross-license 4,371,752 against
IBM's 10,000 patents in the voice messaging area.
One reason for being careful with AT&T is
that, in American patent practice, priority goes, not to the first
to file a patent application as in other countries, but, very often,
to the first to conceive, and the "diligence" with which the idea
was followed up. Makes lots of money for lawyers, but gives
companies with large R&D departments a considerable advantage. The
tic tac toe machine is a case in point.
Some years ago, somebody got a patent for a
machine that played tic tac toe, and hit AT&T for infringement. AT&T
immediately pulled out stacks of lab notebooks showing priority, and
the use of the tic tac toe machine in PR exhibits all over the
country, at World Fairs, in museums, etc., proved diligence. They
just hadn't bothered to patent anything so simple as a tic tac toe
machine, but they sure weren't going to pay royalties to somebody
else for their own toy. They didn't.
Now that IBM owns Rolm, and Rolm and VMX are
the leaders in voice mail in spite of Rolm's somewhat later start,
each with about 25% of the market, IBM has dropped its own voice
mail and is concentrating on Rolm's.
But Wang, another pioneer in the voice mail
business, has elected to pay license fee to VMX. How this relates to
InteCom's use of VMX rather than Wang's system is not known;
probably InteCom had contracted with VMX before Wang bought InteCom
stock. In any event, the situation is not without interest.
But there is one more ring to our present
circus. The June 3 issue of Electronic News reported that Elk
Industries, Inc., of Ft. Lauderdale, filed suit against IBM/Rolm for
infringing Rob Elkins' patent on voice mail, and then, later,
settled out of court for a "six figure sum." The Elkins patent,
4,124,773, issued in November, 1978, is reported to cover digital
voice storage and voice store-and-forward; Mr. Elkins, now 30, got
the idea while in high school. I, for one, would love to see his lab
notebooks, his demonstration of diligence, etc. They must be very
interesting, because Elk Industries has also filed suit against Wang
and DEC, even though IBM/Rolm insist they violated no patent rights
and settled for less than a court case would have cost.
The intrepid Electronic News (9/9/85)
reports that Elk Industries was set up by Mr. Elkins and his two
lawyers "for the purposes of litigation," and says Elk is ready to
go after 40 other companies allegedly infringing its patents. This
would appear to be the second most interesting series of nuisance
suits in modern telecom history. But I'm not sure I'll take Elk
seriously until it takes on AT&T. The tic tac toe machine needs to
be kept in mind but, if Elk runs into any snags, I'll bet Bill
McGowan will be glad to offer a word or two of encouragement.
[
Top ] [ Next ] [
Table of Contents
] |
