Dave, let me answer one of your questions now, and return here tomorrow for the rest. It's been a long day.
In concise terms, silence suppression is noise generation. Aside from this, it has to do with some other significant properties, but that's it, in a nutshell: Noise Generation.
Chinese proverbs and the principles of the Tao undoubtedly pop up when you hear this, right?
In the absence of "side tone" (which is a form of feedback that you normally hear when you speak into a live phone, which resembles a kind of white background noise) the tendency is to blow into the mouth piece to see (hear) if the line is still alive. This may seem like a trivial matter, but believe me when I tell you, it is not.
I once had a TASI system to Paris which was routed over TAT-4 actually fail on me, because the noise generators failed. Read on.
TASI stands for Time Assignment Speech Interpolation. In another era, TASI was used to support 96 talkers on a 36 channel submarine system. Keep in mind here that the 36 submarine channels were already FDM multiplexed.
The system worked on a rather ingenious principle, when you consider that it was conceived in 1939, and only implemented in the early Sixties. It wasn't until they could leverage the lower power consumption characteristics of high-hat transistors that they could get the size AND HEAT DISSIPATION of the machine down to manageable proportions. The vacuum tube equivalent of TASI would have taken up an entire building, and would have required one helluva heat sink.
The one I oversaw in '68 was transistorized and it was actually about the size of a Main Frame, taking up a wing of a skyscraper floor, to give you an idea of how the transistor reduced the space requirements. About 4,000 square feet, in other words, when you included the admin area. Today, it would fit on my desktop, next to my stapler.
How it worked: When a talker on TASI ceased speaking, even momentarily, a channel was rendered inactive and "silent." At that point in time (plus a few milliseconds) it became available for another talker to use, taking advantage of available pause times between syllables and words.
Ordinarily, during silent periods, the line would appear to go "dead." This would have caused talkers to blow into the microphone to test its liveliness, i.e., the side tone. If talkers did this, however, they would activate the gates to idle channels, seizing them, in effect, and shutting those channels out to legitimate talkers who were in need of one.
To avert this, noise generators were used. Nose was directed at those talker trunks whose talkers had ceased speaking. This would satisfy them, in a spoofing way, causing them to believe that they had a live channel. So, in this fashion, when a talker ceased talking, they would continue to hear "side tone" and background noise.
On overseas circuit's background noise was not merely white noise, however, as anyone who has ever spoken on a pre-fiber connection to Europe will attest. It was mixed with an eery combination of background dial-pulsing tones, single frequency signaling tones, cross talk, and a complete symphony of twilight zone -like effects that will never be duplicated again.
So, the noise had to be significantly rich and deep, in order to spoof the experienced transatlantic talker.
The work-around? Noise generators. Or, as they are now called, silence suppressors.
Similar dynamics are now at play on VoIP nets, since silence would elicit the same kinds of blow-in-the-mouthpiece response, which would seize channels and unnecessarily utilize precious bandwidth and other VoIP/ruoting resources.
----
While working the overseas board one day in '69, the primary noise generator card went south while the backup was being "routine tested." The Paris TASI System thus also went south... as did some 96 unhappy talkers.
----
A related (maybe opposite?) issue that is going to cause a great deal of havoc on VoIP systems in the near future will be: Music On Hold.
Think about it.
Regards, Frank_C. |
