Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!uwm.edu!ux1.cso.uiuc.edu!usenet From: mcdonald@aries.scs.uiuc.edu (Doug McDonald) Newsgroups: comp.compression Subject: Re: Sound compression Message-ID: <1991May24.144519.4129@ux1.cso.uiuc.edu> Date: 24 May 91 14:45:19 GMT References: <1991May24.083409.20528@looking.on.ca> <91May15.140250edt.750@neuron.ai.toronto.edu> <105539@sgi.sgi.com> <1991May23.123202.18335@cc.tut.fi> <1991May23.215341.7836@nntp-server.caltech.edu> Sender: usenet@ux1.cso.uiuc.edu (News) Organization: University of Illinois at Urbana Lines: 45 In article <1991May24.083409.20528@looking.on.ca> brad@looking.on.ca (Brad Templeton) writes: >You would think sound should be more redundant than that. It appeared that sound is not redundant at all. All attempts to compress digitized sounds have failed in the sense that critical ears can hear the difference. Even the 16 bits of present CDs is not completely adequate. 19 bits probably is, and the 44 kHz sampling rate probably is - given really good pre and post analog filtration. It is true that you can do some really radical things to audio in the analog domain -- like screwing up the phase-versus-frequency linearity -- without audible effect, but that does not reduce the information content of the signal. Of course this does not apply to synthetic, electronically generated sounds, as, of course, you can compress the hell out of, for example, a signal consisting of two sine waves. >I had heard >plans for 2.5inch CDs with over 2 hours of sound in the past, I guess >those got scrapped. At 2 hours, they would begin to have trouble >figuring what to put on it. Even today people are more ruluctant to >buy 30 minute CDs when 70 minute CDs are abundant. And of course long >double-albums get to charge double, when we all know the cost of the CD >is peanuts. > >I have heard of some people doing 1 meg/minute for supposed hi-fi. >Do they lie? Yes. There is a LOT of sloppy pseudo-scrience crap in the audio world, and the so-called golden ears often fail to hear certain non-information-losing transformations, but the attempts to compress music using informatrion-losing methods have all been audible. Doug MCDonald