Newsgroups: comp.multimedia
Path: utzoo!utgpu!cunews!dgbt!ted
From: ted@dgbt.doc.ca (Ted Grusec)
Subject: Re: Pitch and Colour
Message-ID: <1991Apr12.001101.8927@dgbt.doc.ca>
Organization: The Communications Research Centre, Ottawa, Canada
References: <1991Apr10.031342.27656@dgbt.doc.ca> <1991Apr11.132547.16696@cbnewsh.att.com>
Distribution: na
Date: Fri, 12 Apr 91 00:11:01 GMT

I was pulling figures out of imperfect memory and got them slightly
wrong.  Maximum sensitivity is at about 4K hertz, and real-life speech
is from about 200 herz to about 6K hertz. You're probably right about
the telephone. However, the telephone is not a hi-fi instrument. The
main point I was making, however, is that you don't need 10 octaves
for speech, as the person I was replying to claimed. It's more like 5
octaves in the full fidelity situation, and much less than that over
the telephone. Even within the restrictions of the telephone,
intelligibility is good, and, more than that, speaker identifiability
is also good.