jane@hpuxa.ircc.ohio-state.edu (Jane M. Fraser) (09/07/90)
I've been reading the comments on TASI with great interest since I've heard about it since I was a toddler, mumble years ago. My father, John M. Fraser, was one of the lead people in the systems engineering for TASI. He retired from Bell Labs in 1972, after 35 years with them. I finally got ahold of him last night and asked him about TASI. I'm sure I didn't get all this information exactly right. I'm not a telephone engineer (I am an industrial engineer). Anyway, here goes. Someone already gave the correct acronym: Time Assigned Speech Interpolation. Dad said the basic idea was described in the Bell Systems Technical Journal very early; he thinks maybe in the 30s, but that article ended by saying that, of course, relays can't switch fast enough. This changed, again of course, with the invention of the transistor. However, because of that early article, Bell Labs had no patent on the technique. The first transatlantic phone cable was finished in 1955 (Dad was a Bell Labs representative on the ship, the Long Lines, that laid the cable, and in Oban, Scotland, where the cable came ashore). He knows they got started on TASI right away and guesses it was implemented in the early 60s. He thinks the first system would have been installed about 1962. Dad said they did some measurements on the percent of time a speaker actually talks, using conversation between operators and found each one spoke, on average, 39.5% of the time. He rounded it to 40% and this became gospel. The inverse of .4, gives 2.5, which is the maximum compression theoretically possible. They did make TASI work with 36 circuits (putting 72 speakers on it), but the statistics improve considerably with more lines. It is currently used on telephone cables (the Japanese have used it quite a bit), but is not used on satellite circuits, as far as he knows, because of the considerable clipping that already exists. (After retirement from Bell Labs, Dad worked for Hughes Communication Satellites as a telephony expert and did much of the systems engineering for Palapa, the communication satellite for Indonesia.) It is used even on land lines when there is an emergency and fewer circuits are available. He said there is (or at least was, when he was involved) no clipping of final sounds, only initial sounds. He said the system was engineered to give only .1% clipping on initial sounds, which is acceptable to most listeners. He said most speakers can't detect when they are on a TASI'd line, although he said my mother always could. Dad said all the technical information was published in old IEEE journals - probably when it was still the IRE, probably in the Transactions. It would all also be in old issues of the Bell Systems Technical Journal. He (modestly) said one of the best articles is ``Engineering Aspects of TASI," by K. Bullington and J.M. Fraser, BSTJ, volume 38, number 2, March 1959. If you want more details, Dad said he'd be happy to chat. Call him at 619-239-2620 (San Diego). He has lots of great stories. Ask him about the time Prince Phillip's boat anchored on the cable in Oban harbor. Jane Fraser