smb@ulysses.UUCP (Steven Bellovin) (05/10/85)
[in which the fire extinguisher was held at the ready, and the cat to jj's bronchial tubes] In response to (un?)popular demand (and not a few flames), we decided to compare some CD players using non-Bose 901s. In particular, we used a Carver Receiver, a pair of DCM Time Windows (I'm told there are submodels, but I don't know which I have; there are no indications on the speakers themselves. They were purchased less than one year ago. --smb), three CD players (jj's (in)famous Magnacrock 2020, a (different) CDP-101, and a Technics SL-P7) taken two at a time, and Kalypso, the famous double-wide audio cat (note: this is a different cat from either of the two used in the previous test). Additional equipment used included some Molson's Golden, one Prior Double Dark for jj (he's picky; you know these old sound people) a 1965 dime from the Philadelphia mint (see below), A Southern Season's "Irish Coffee" blend of coffee beans ground in a Braun grinder and brewed in a Krups coffee machine, some popcorn, and some tortilla chips with hot sauce. (The cat finished off a bowl of tortilla chips incautiously left on the floor, which is one reason she's a double-wide.) The receiver and the speakers are in different rooms; from the listening area, it is not possible to see the receiver or to observe the motions of someone switching between two different inputs. Neither the munchies nor the cat had any fixed location. The active participants were Steve Bellovin (ulysses!smb) Lynda Feng (hound!llfe) Dick Grantges (hound!rfg) Jim Johnston (alice!jj) Phil Karn (bellcore!karn) Andy Koenig (research!ark, alice!ark) Marty Shannon (eagle!mjs) As before, we matched signal levels from the CD players to within 1%; we also took care that the voltages at the speaker terminals matched. In the process, we learned that the `1 kHz' tone on the Denon test CD is about a semitone higher than the `1 kHz' tone on the Elektra test CD. We compared several different pieces: Scheherazade (Kondrashin, Philips 400 021-2; recommended to us by Craig Dory, who was invited but could not attend), Bach's Chromatic Fantasy (Pinnock, Archiv 413 638-2), Beethoven's "Emperor Concerto" (Serkin/Ozawa, Telarc CD-80061-5), and Holst's "The Planets" (Maazel, CBS MK 37249). Results were slightly more ambiguous than during the first test. During "Scheherazade", jj observed some extremely minor differences in imaging; twice, he guessed which CD player (the Magnavox or the Sony) was in use (twice right, once wrong (some difference, right! Actually, it's my feeling that there is a very small difference, but the only way to tell would be a "is this different" test under much more tightly controlled situations. Actual identification is out. --jj)). Smb sometimes thought he heard it, too, but was unable to consistently identify the player. Smb suggested that the switching transient itself was audible, which might have created a subconscious desire to hear *something*. During the Beethoven, jj again observed some differences; smb, and to a lesser extent karn and ark, heard what might have been some difference. No one was completely convinced that the effect was real. Ark guessed twice at which player was in use (wrong both times). Neither mjs nor rfg heard anything at all different. No one heard any differences during the Bach piece except possibly the cat, since she took a swipe at jj while he was switching inputs. (To be sure, she was listening to a pair of Allison IVs at the time.) Perhaps she had an opinion on which player she preferred, and did not want him to change. The piece was closely miked, though, which could interfere with imaging differences. Everyone remarked on how many notes there are in the Chromatic Fantasy. At this point the cat again took a hand (a paw?) -- allergic reactions forced mjs to leave, taking his CDP-101 with him. We connected karn's SL-P7 instead, it having analog filtering as well. This time, we performed a double-blind test with jj as the experimental subject. A list of 16 random bits was prepared by smb using the afore-mentioned dime. We gave jj a chance to listen to each player until he was ready to begin the actual test, telling him which player was which each time. Once the test began, jj was presented with 16 segments played by one player or the other, as determined by the random bits. After listening to each segment, jj waved his hand when he wanted to begin the next segment. Smb relayed this signal to ark, who operated the controls; ark was the only one who knew which player was on at any one time. (Rfg and karn were in the listening room with jj, ark could not see anyone but smb, and smb could not see the controls.) To disguise null shifts, all switches were made by going first to an unused input source for a fraction of a second, and then to the desired player. JJ recorded his impression of which player was on, arbitrarily calling the first selection `A'. No attempt was made to record which player he thought was `A' and which was `B'. When the test was over, his record was compared with the input table. The results: *random*. That is, of the 16 answers, 8 were consistent with the first choice; 8 were not. (This being in line with JJ's impressions of the results before scoring.) We ended the evening by actually listening to some music (rather than equipment), and had some fun observing just how good the imaging was on some organ pieces. We also tried a sound effects track, which terribly confused the cat. Ark commented that he thought the Time Windows sounded similar to his 901's... JJ disagreed, very quietly. Conclusions: our results this time won't satisfy anyone. The most carefully controlled part of the testing failed to establish anything; on the other hand, time and cat allergies (jj and mjs) prevented us from doing another double-blind test using the second available CDP-101 and Scheherazade. We will probably go through this twice more, using rfg's AR-9s and jj's homebuilt speakers. Unfortunately, time and schedule constraints will delay the next round until July or thereabouts. We invite others to conduct similar tests on their own. If you do, we cannot overemphasize the importance of level matching (with a voltmeter!) and synchronization. Other observations: (a) such differences as were noticed were quite small in magnitude; to smb's thinking at least, neither player could be called "better" or "worse". They would be dwarfed by the difference between any sets of good speakers. (b) the strength of the psychological desire to hear *something* cannot be overstated. Given how noticeable the switching transient was, it would be very easy to imagine differences when none were actually present. (By comparison, the switching transient last time was far less noticeable. This may be because the Heath preamp used has a mechanical switch, whereas the Carver Receiver has an electronic switch.) (c) the differences in the human factors of the different players was far greater than any sonic difference; even if you can hear the difference when listening at your peak, you would do well to give greater weight to the controls. (d) it *is* possible to destroy a CD via relatively small scratches... (*sigh*) --Steve Bellovin Andy Koenig Lynda Feng Dick Grantges Jim Johnston Phil Karn Marty Shannon
herbie@watdcsu.UUCP (Herb Chong [DCS]) (05/11/85)
In article <970@ulysses.UUCP> smb@ulysses.UUCP (Steven Bellovin) writes: ....details of testing and results deleted.... >We invite others to conduct similar tests on their own. >If you do, we cannot overemphasize the importance of level matching >(with a voltmeter!) and synchronization. for others doing this test, the voltmeter must have adequate and known frequency response. the critical 1 kHz test tone should be well within the voltmeter's bandwidth and the two players adjusted to within 0.1 and preferably 0.05 dB. if possible, electronic switching should be used with a HIGH quality relay circuit. >Other observations: (a) such differences as were noticed were quite >small in magnitude; to smb's thinking at least, neither player could be >called "better" or "worse". They would be dwarfed by the difference >between any sets of good speakers. (b) the strength of the psychological >desire to hear *something* cannot be overstated. Given how noticeable >the switching transient was, it would be very easy to imagine differences >when none were actually present. (By comparison, the switching >transient last time was far less noticeable. This may be because the >Heath preamp used has a mechanical switch, whereas the Carver Receiver >has an electronic switch.) (c) the differences in the human factors of >the different players was far greater than any sonic difference; even >if you can hear the difference when listening at your peak, you would >do well to give greater weight to the controls. (d) it *is* possible to >destroy a CD via relatively small scratches... (*sigh*) most of these results are well known in psychoacoustics papers by various people. a big influence on whether a difference is heard or not is the expectations of the listener. the more paid for the piece of equipment, the greater the desire to hear superiorities, imagined or not. frankly, i expect that about the only audible difference that is significant between CD players is minor variations in frequency response. this is not to say that there are a lot of theoretical advantages to oversampling, but to say that one technique is audibly better than another when both are so good is not very important. > --Steve Bellovin > Andy Koenig > Lynda Feng > Dick Grantges > Jim Johnston > Phil Karn > Marty Shannon another test with a bona fide high end, cost-no-object system under double blind and controlled conditions and known "golden ears" would help further these conclusions. AudioScene Canada conducted such a test about three years back to see whether there was a difference between MC and MM cartridges. several highly respected cartridges of each type selected by the underground audiophile community were tested with specially selected electronics and equipment under very tightly controlled conditions. the NRC audio research studios were used for the comparison. many Toronto and Ottawa area "golden ears" were invited and did these tests individually and as a group. no discussion or conversation of any kind was allowed until the tests were complete. the results were "ego shattering" as one of the testee's said later. out of the panel of 17 (or was it 16?), only one could distinguish between any MM and any MC cartridge in a consistent manner. if people are interested, i can post more detailed results and the summary. Herb Chong... I'm user-friendly -- I don't byte, I nybble.... UUCP: {decvax|utzoo|ihnp4|allegra|clyde}!watmath!water!watdcsu!herbie CSNET: herbie%watdcsu@waterloo.csnet ARPA: herbie%watdcsu%waterloo.csnet@csnet-relay.arpa NETNORTH, BITNET, EARN: herbie@watdcs, herbie@watdcsu