geek@media-lab.media.mit.edu (Chris Schmandt) (09/03/90)
A number of conversations can share a significantly smaller number of circuits by freeing the circuit associated with a conversation during silent periods, and re-assigning one when speech starts. I recollect such a scheme being referred to as "TASI", but can't find this term in the online glossaries. Can anyone enlighten as to what it stands for? Is there a standard number for how much "circuit compression" it provides? Thanks in advance, chris
hes@ccvr1.cc.ncsu.edu (Henry E. Schaffer) (09/04/90)
[in a section on TDM] "... to TDM several speech channels by taking advantage of pauses between words and statements. Utilizing this principle, a Time Assignment Speech Interpolation (TASI) system is used on many overseas channels to effectively increase the capacity of the channels. Again, to avoid problems with more simultaneous talkers than available channels, the number of talkers and channels should be rather large (100 or so). Obviously, the switching in such a scheme must be very rapid, and the resulting complex equipment is not attractive except for use on expensive channels such as overseas applications." From Transmission Systems for Communications, 4ed, 1970, Bell Labs. Since this reference is 20 years old, the costs and tradeoffs have changed. henry schaffer n c state univ
goldstein@carafe.enet.dec.com (Fred R. Goldstein) (09/05/90)
In article <11667@accuvax.nwu.edu>, geek@media-lab.media.mit.edu (Chris Schmandt) writes... >A number of conversations can share a significantly smaller number of >circuits by freeing the circuit associated with a conversation during >silent periods, and re-assigning one when speech starts. I recollect >such a scheme being referred to as "TASI", but can't find this term in >the online glossaries. > Can anyone enlighten as to what it stands for? Is there a standard >number for how much "circuit compression" it provides? Time Assignment Speech Interpolation is actually a form of packetized voice transmission. It was first used in the early 1970s on undersea cables. The idea is to take small chunks of audio, run them through a level detector, and only send them if they aren't "silent". Then each one can be prefaced with a channel header so the receving end knows which channel is getting which call over the trunk pool. Typically you get nearly 2:1 compression with voice. While AT&T's large TASI systems came first, many private networks used the Storage Technology COM-II TASI about a decade ago. (We did here at Digital.) But analog TASI is obsolete. Newer Digital TASI, now called DSI, systems generally run over T1 carrier. The StrataCom IPX is such a box; it sends 192-bit frames with a 24-bit header and 168-bit audio payload. It's nearly toll-quality even with silence suppression enabled. Another example is AT&T's Integrated Access Cross-Connect Switch (IACS), which uses frame relay-based DSI. It's widely used on AT&T's international network. You may have used it without knowing. Typical 2:1 compression ratios are being impacted by fax. Since fax (like dial-up data) sends a constant audio signal, it can't be compressed. We had the same problem on our TASI years ago, and actually routed a lot of dial-up data around it. Fred R. Goldstein goldstein@carafe.enet.dec.com or goldstein@delni.enet.dec.com voice: +1 508 486 7388 opinions are mine alone; sharing requires permission
BRUCE@ccavax.camb.com (Barton F. Bruce) (09/05/90)
> circuits by freeing the circuit associated with a conversation during > such a scheme being referred to as "TASI", but can't find this term in TASI stands for something like: "Time Assignment Speach Interpolation" if I remember correctly. It was expensive to implement, so was mostly used on transatlantic cables. The old analog versions have given way to the modern digital versions marketed to folks with large global nets and big budgets. Not only is no bandwidth given you in quiet periods, but also your speech may be slightly further delayed if buffers instantaneously backup beyong the capacity of the pipe to carry it all. When you can't see the lips speaking, how are you to know that that pause was more than just satelite delay? Also beginnings and ends of sylabyls may get slightly clipped, and the human ear, wonderful thing that it is, never misses anything. Republic Telecom can easily give you 40 voice channels on 5 x 56kb, with some provision for handling fools who try to sneak modem traffic across where it does not belong. That is a small size box for them. Folks with more modest budgets may want to check Pacific Communication Sciences, Inc, too, if you want clever voice compression, but not traditional TASI.
sp@questor.wimsey.bc.ca (Steve Pershing) (09/05/90)
TASI was (I think) invented by Bell Labs at least 15 years ago. It stands for Time Assignment Speech Interpolation. The original purpose was to provide more voice circuits on overseas cables. Since the actual time that a voice is really speaking is rather small, the original TASI allowed for somewhere around a 5:1 increase in conversations. Internet: sp@questor.wimsey.bc.ca | POST: 1027 Davie Street, Box 486 Phones: Voice/FAX: +1 604 682-6659 | Vancouver, British Columbia Data/BBS: +1 604 681-0670 | Canada V6E 4L2