This is a purely subjective test and as such there is no right and wrong answer. Pure listening impressions only.
If there is sufficient interest we could can run a hidden poll on this.
IMPORTANT Please don't post any listening impressions on the thread. As we have found before, that just influences and skews the outcome. You can PM me with your choice and then after a few days the result can be made public.
How the test works. The following three groups of files (which are all .wav) are of the same test. The only difference is the music choice. The first two sets are of Genesis and the last Joni Mitchell. Neither are particularly outstanding, and date from the early and late 1970's. The Genesis in particular has a reputation for sounding a little "thin" and "glassy".
(Also the change in what you are listening to is not consistent between groups. For example if you preferred Beta over Alpha then it doesn't follow that you would prefer Lambda over Kappa or Pi over Omicron).
If you ABX these files you may well consistently pick one over the other (which would be great) but then the real question is which you prefer and why.
Here are the Dropbox links...
Group 1
ALPHA
BETA
Group 2
KAPPA
LAMBDA
Group 3
OMICRON
Pi
If there is sufficient interest we could can run a hidden poll on this.
IMPORTANT Please don't post any listening impressions on the thread. As we have found before, that just influences and skews the outcome. You can PM me with your choice and then after a few days the result can be made public.
How the test works. The following three groups of files (which are all .wav) are of the same test. The only difference is the music choice. The first two sets are of Genesis and the last Joni Mitchell. Neither are particularly outstanding, and date from the early and late 1970's. The Genesis in particular has a reputation for sounding a little "thin" and "glassy".
(Also the change in what you are listening to is not consistent between groups. For example if you preferred Beta over Alpha then it doesn't follow that you would prefer Lambda over Kappa or Pi over Omicron).
If you ABX these files you may well consistently pick one over the other (which would be great) but then the real question is which you prefer and why.
Here are the Dropbox links...
Group 1
ALPHA
BETA
Group 2
KAPPA
LAMBDA
Group 3
OMICRON
Pi
I don't believe this will spoil anything to do with subjective listening impressions:
OMICRON/Pi are loaded into Cool Edit Pro, a waveform peak about 7.5sec into track was focused on. Based on this region tracks where realigned by 121 samples. Next sample rates where changed up 8x. At new sample rate and additional track shift of 1 sample was made. Tracks converted back to fs 44.1kHz. Amplitudes of waveforms in 2kHz region of spectrum observed, and amplitude adjustment of 0.01 dB applied. Sample differences of tracks are made, and spectrum referenced to alignment point observed:
OMICRON/Pi are loaded into Cool Edit Pro, a waveform peak about 7.5sec into track was focused on. Based on this region tracks where realigned by 121 samples. Next sample rates where changed up 8x. At new sample rate and additional track shift of 1 sample was made. Tracks converted back to fs 44.1kHz. Amplitudes of waveforms in 2kHz region of spectrum observed, and amplitude adjustment of 0.01 dB applied. Sample differences of tracks are made, and spectrum referenced to alignment point observed:
Hmm... don't know what to say to that tbh. I've never used Cool Edit so I'm not really sure what I'm looking at if I'm honest 
So starting at the left we have four ? traces. Two blue and two green each overlaid (apart from slight differences). What do they represent ? What are the green and what are the blue showing ?
(Sorry for the dumb questions I need it explaining what I'm looking at)
So starting at the left we have four ? traces. Two blue and two green each overlaid (apart from slight differences). What do they represent ? What are the green and what are the blue showing ?
(Sorry for the dumb questions
I need it explaining what I'm looking at)
Thank you
and a very happy new year to you too.Come on folks, the more that listen and the better.
Mooly,
This green traces are L/R spectrum of OMICRON/Pi at 7.46sec. Blue traces are spectrum of difference of two files at 7.46sec. Files as posted are not aligned. shifts in time and amplitude were applied to better align them to reduce residual seen as blue traces.
Seems that before a subjective opinion can be given about the sound of both conditions, listener needs to demonstrate ability to hear identifiable difference between the recordings.
This green traces are L/R spectrum of OMICRON/Pi at 7.46sec. Blue traces are spectrum of difference of two files at 7.46sec. Files as posted are not aligned. shifts in time and amplitude were applied to better align them to reduce residual seen as blue traces.
Seems that before a subjective opinion can be given about the sound of both conditions, listener needs to demonstrate ability to hear identifiable difference between the recordings.
Thanks for the explanation... I'll have to think about that
As to identifiable differences, yes I know for certain that I can pick one over the other nearly every time. This morning I tried an ABX on Kappa and Lambda and managed the following. Concentration waned at the end and I fluffed the last result only managing 14/15
But the only question with this test is which you prefer and why.
As to identifiable differences, yes I know for certain that I can pick one over the other nearly every time. This morning I tried an ABX on Kappa and Lambda and managed the following. Concentration waned at the end and I fluffed the last result only managing 14/15
But the only question with this test is which you prefer and why.
Attachments
Kappa lags Lambda by 92 samples, 2 milliseconds.
You appear to be doing ABX keeping playback position when switching tracks. This has effect of jumping forward/backwards 2 milliseconds when changing tracks. This is readily detectable when performing ABX, and needs to be eliminated as potential basis for discrimination.
Delete 92 samples from start of Kappa and see how it effects your ABX performance.
You appear to be doing ABX keeping playback position when switching tracks. This has effect of jumping forward/backwards 2 milliseconds when changing tracks. This is readily detectable when performing ABX, and needs to be eliminated as potential basis for discrimination.
Delete 92 samples from start of Kappa and see how it effects your ABX performance.
Well, very little interest in this thread (and many thanks to those that did participate) so I think I may as well show what the set up was.
The test set up was a Sony CDP790 THIS ONE
Here is the circuitry of the output stages.
The 5532 was a JRC5532 of single in line style. That made trying modifications difficult but I made a couple of little adapter PCB's to allow DIP devices to be fitted. Notice the supplies to the opamp, just -/+5 volts. That's low.
Now I wanted to try FET opamps in this location and chose the TLE2072 as a suitable candidate. The data sheet says "fully specified for operation at -/+5 volts". The results however were slightly unexpected with a huge increase in distortion as levels approached 0db level. What was surprising was the distribution of this "level dependent" distortion which you can see here.
So here we have the distortion as level increases. Ignore the HF artifacts at 18 to 20kHz as they are generated by the sound card.
-10db
-6db
-3db
-1db
0db
So as you can see, we have a "classic" distribution of distortion with very high second harmonic as level tops out toward 0db.
I was intrigued whether anyone preferred this particular "sound". The two results I received were very interesting.
fas42 never even blinked
Beta
Kappa
Pi
were the preferred choices. The comments included. Quote "Smoother, less "hifi", less edge, the usual thing - the other version just started to bug me more the longer I listened ...".
So fas42 preferred the totally original set up of the 5532.
The other result (thanks BYRTT) preferred Beta, Kappa and Omicron but with special mention of the acoustic guitar on Pi coming across particularly well.
I think the Omicron and Pi tracks might have a higher overall level then the others (the masters on the CD... I left all settings the same between tracks) and that would push the distortion ever higher.
Many thanks guys. I'll leave the files available for a little while in case anyone wants to try these.
The test set up was a Sony CDP790 THIS ONE
Here is the circuitry of the output stages.
The 5532 was a JRC5532 of single in line style. That made trying modifications difficult but I made a couple of little adapter PCB's to allow DIP devices to be fitted. Notice the supplies to the opamp, just -/+5 volts. That's low.
Now I wanted to try FET opamps in this location and chose the TLE2072 as a suitable candidate. The data sheet says "fully specified for operation at -/+5 volts". The results however were slightly unexpected with a huge increase in distortion as levels approached 0db level. What was surprising was the distribution of this "level dependent" distortion which you can see here.
So here we have the distortion as level increases. Ignore the HF artifacts at 18 to 20kHz as they are generated by the sound card.
-10db
-6db
-3db
-1db
0db
So as you can see, we have a "classic" distribution of distortion with very high second harmonic as level tops out toward 0db.
I was intrigued whether anyone preferred this particular "sound". The two results I received were very interesting.
fas42 never even blinked
Beta
Kappa
Pi
were the preferred choices. The comments included. Quote "Smoother, less "hifi", less edge, the usual thing - the other version just started to bug me more the longer I listened ...".
So fas42 preferred the totally original set up of the 5532.
The other result (thanks BYRTT) preferred Beta, Kappa and Omicron but with special mention of the acoustic guitar on Pi coming across particularly well.
I think the Omicron and Pi tracks might have a higher overall level then the others (the masters on the CD... I left all settings the same between tracks) and that would push the distortion ever higher.
Many thanks guys. I'll leave the files available for a little while in case anyone wants to try these.
Its bang on the CD standard, 5.6 volts pk/pk (2volts rms). The test disc is the Philips "audio signals disc 1" pictured in the link to the player. The original 5532 is totally clean.
It seems to be just the low supply causing this, I was very surprised. The layout of the adapter is straightforward but I might run some parallel test with the TLE2072 in a circuit on low rails just to see.
Hi Pavel. An interesting observation. Even at -6db some distortion residual is still observable. As previous tests have shown, some can discern minute differences between circuits, differences that almost defy measurement even with the best of set ups. I'd bet there is much going on even below these levels if only I had the means to measure it. At the higher output levels the distortion could be observed on a scope as (to me) classic asymmetric 2nd harmonic. There was no visible suggestion of it approaching a hard clip limit.
Is it
There is a genuine change of circuit between the pm files. Rails are the same and its not using a TLE2072 either
You might like to try Arta
ARTA Software
It is for free, very well designed, with much higher resolution than Audacity. Audacity is nice for music file handling, but close to useless for spectral analysis and distortion measurement. With Arta, you will know much more about opamps and soundcards.
ARTA Software
It is for free, very well designed, with much higher resolution than Audacity. Audacity is nice for music file handling, but close to useless for spectral analysis and distortion measurement. With Arta, you will know much more about opamps and soundcards.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.