What I am saying is that when you cancel two analog recordings by substracting them, 99 % of the difference comes from erratic time drift. Extracting a useful information from the result is like extracting the cosmic background radiation from the static on the TV.
Imagine that between the two recordings, you have a lenght difference of 1/44100th of a second. That is still an accuracy better than 10^-7 over a 5 minutes recording.
In the worst case, a 0 dB 22.05 kHz sine, the amplitude error will reach the complete 65536 steps of the 16 bits recording, since you get the next sample instead of the good one !
In order to cancel two signals with an accuracy of 16 bits over a 5 minutes recording, you need a timing accuracy of 10^-12 !
Therefore signal cancellation is not a practical way to detect any analog difference.
I know minute differences can be detected by analysis. I even detected differences between the direction of interconnects with the RMAA software, with cables that have no direction ! The 50 Hz hum was just differently picked, because I did not exactly replace the cables in the exact same orientation between the two recordings.
It is perfectly possible to measure the differences between two interconnect. They are mainly frequency response at 20 kHz (differences of about 0.01 dB), and 50 Hz Hum. IMD and THD are null, or at least completely masked by the ones of the DAC and ADC.
Both are way, way below audible thresholds.
Face it, there are NO measurable parameters in interconnects that have any relationship with their "sound". The only way out is blind testing.
As soon as some people who can reliably hear differences between interconnects begin to investigate this way, we can learn new things. So far, most testers were people who usually don't hear differences, or very little ones.
Grand X, in France, recently did some blind testing with a good perception of differences between cables. The test failed. His conclusions so far is that he doesn't understand why the differences change or disappear during the test, and he wishes to go on until he understands why.
One of the most interesting part in this ABX was that differences were clearly heard, but not when the cables were changed, even when the listeners knew the cable was not changed !
For example, after the presentation of X in the 6th trial, one listener asked to listen to the reference, then to the same random cable X again. Several people had the very clear feeling that the sound was no the same at all ! They changed their ABX answer for this trial, but were completely puzzled, because they did not understand why this X cable sounded like A (or B) when first presented, and like B (or A) when presented again.
That's an example of phenomenon that a listener can investigate in order to qualify the sound of cables. There are things to learn here.
Imagine that between the two recordings, you have a lenght difference of 1/44100th of a second. That is still an accuracy better than 10^-7 over a 5 minutes recording.
In the worst case, a 0 dB 22.05 kHz sine, the amplitude error will reach the complete 65536 steps of the 16 bits recording, since you get the next sample instead of the good one !
In order to cancel two signals with an accuracy of 16 bits over a 5 minutes recording, you need a timing accuracy of 10^-12 !
Therefore signal cancellation is not a practical way to detect any analog difference.
I know minute differences can be detected by analysis. I even detected differences between the direction of interconnects with the RMAA software, with cables that have no direction ! The 50 Hz hum was just differently picked, because I did not exactly replace the cables in the exact same orientation between the two recordings.
It is perfectly possible to measure the differences between two interconnect. They are mainly frequency response at 20 kHz (differences of about 0.01 dB), and 50 Hz Hum. IMD and THD are null, or at least completely masked by the ones of the DAC and ADC.
Both are way, way below audible thresholds.
Face it, there are NO measurable parameters in interconnects that have any relationship with their "sound". The only way out is blind testing.
As soon as some people who can reliably hear differences between interconnects begin to investigate this way, we can learn new things. So far, most testers were people who usually don't hear differences, or very little ones.
Grand X, in France, recently did some blind testing with a good perception of differences between cables. The test failed. His conclusions so far is that he doesn't understand why the differences change or disappear during the test, and he wishes to go on until he understands why.
One of the most interesting part in this ABX was that differences were clearly heard, but not when the cables were changed, even when the listeners knew the cable was not changed !
For example, after the presentation of X in the 6th trial, one listener asked to listen to the reference, then to the same random cable X again. Several people had the very clear feeling that the sound was no the same at all ! They changed their ABX answer for this trial, but were completely puzzled, because they did not understand why this X cable sounded like A (or B) when first presented, and like B (or A) when presented again.
That's an example of phenomenon that a listener can investigate in order to qualify the sound of cables. There are things to learn here.
