ted@dgbt.doc.ca (Ted Grusec) (06/24/91)
I am looking for some advice and information. I am involved in
subjective testing of various coding algorithms to see whether
listeners can hear differences in pieces of audio (mostly
music) between original ("CD quality") and coded versions.
All of this is pertinent to the next generation audio broadcasting
(and recording) systems, which will be fully digital. The coding schemes
all throw away information which is masked because of the
characteristics of the human auditory/cognitive system, and therefore,
theoretically, not heard. The algorithms perform the computations
to determine what can be thrown away at around 1000 times/sec. The
reasons for throwing the information away is to conserve bandwidth
in the broadcast situation. Some of you may have heard of the
forthcoming Phillips DCC (Digital Compact Cassette) system which uses
one of these schemes and promises to displace the analog cassette
in the consumer arena within a year or so.
My concern is as follows. While the best of these systems perform
remarkably well (i.e. even the most critical listeners can't hear
differences), the audio materials used in the tests are all bits and
pieces of traditional music (including electronic but of the type that
closely mimics traditonal timbres). I wonder whether current and future
developments in computer music might tax these coding schemes to a
point where they are no longer transparent to the listener. It would
be a tragedy if international decisions on the next generation
broadcasting precluded the undistorted transmission of computer music.
This music is hard enough to get performed at best. We don't need
to hamper public exposure by making it difficult to broadcast such
music without adding distortion.
The kinds of timbres which give these algorithms that hardest time are
very sharp attacks and "brilliant" overtones. Can anyone suggest
commercially available CDs which might contain elements that should be
examined in transparency tests? Ten to 20 sec. bits of audio are all
that are needed. Or, does anyone have original materials which they
could provide in digital format and which might be used in subjective
tests?
--
==========================================================================
Ted Grusec Communications Research Centre (Govt. of Canada, DOC)
3701 Carling Ave., Ottawa, Ont.K2H 8S2, CANADA Internet: ted@dgbt.doc.ca
Compuserve: 73607,1576 (613) 998 2762 Fax (613) 993 8657