Well, now we have given people a taste of how we design audio equipment, and what we find important and interesting. It is different from just measurement, yet measurement is important, it is different from just liking something, as it may not be something 'universal' that we can apply to other audio equipment. However, it is about keeping an 'open mind' to what works, usually offering some sort of 'hypothesis' as to why we are hearing what we are hearing, in order to keep track of it for listening to something similar on other occasions.
We don't try to make a 'mystery' of it, but we don't deny what our ears hear either. AND we don't need any ABX tests to convince us of anything, one way or the other.
We don't try to make a 'mystery' of it, but we don't deny what our ears hear either. AND we don't need any ABX tests to convince us of anything, one way or the other.
Got a few more replies on the teaser files. One member was brave enough to come out and admitted that he couldn't sort the files, but with logical reasoning came pretty close to what the processing was that was done.
For those of you that still are afraid to really trust your ears
I posted another file (filename: '1000') that has the same effect but now 1000 times as strong. Don't worry, I can virtually guarantee that you can hear this one.
Found through the same link: Golden Ear Special
jan didden
For those of you that still are afraid to really trust your ears
I posted another file (filename: '1000') that has the same effect but now 1000 times as strong. Don't worry, I can virtually guarantee that you can hear this one.Found through the same link: Golden Ear Special
jan didden
How did the resistor know the needle was put in the groove? I could care less about their credentials but once you eliminate (possibly) carbon comp and 0203 SMT I find these claims exraordinary. The surface noise is an order of magnitude or so higher especially at the low end.
I'm sure Dr. H has a fuzzy theory for this.
Last edited:
Just curious, why didn't you provide a seventh file, a known non-processed file to use as the reference?
se
I think he meant that the surface noise sounded different - sort of like using pink noise as a test signal.
Scott,
It is not the resistor noise he is referring too. It is the noise inherent in the record. That is a wideband noise source. The issue is do resistors in some way influence how that is reproduced by the following stages?
As an example consider two identical resistors but one with more lead inductance. That one will allow more HF energy into the following equipment which even if above the audio band will still cause IM distortions that can drop into the audio band.
So we are not talking about excess noise but rather the RLC variations and possible microdiodes that create bandwidth changes or other distortions.
Well, you have three references! Just got word from Malcolm (he's vacationing is Europe) that the groups are three unprocessed files (the references!) - and three files that have been run through an all-pass to simulate the phase effects of a 4th order LR. Within each group the three files are absolutely identical. No cheating or traps or whatever.
Note also that strictly you don't need to identify the unprocessed ones, just identify the two groups, meaning: can you hear the phase effects of a 4th order LR. The '1000' file might give you a clue what you should search for.
But Malcolm is still holding off as to what the correct grouping is
BTW Those who have read my interview with Siegfried Linkwitz may remember that in 1974, when Siegfried was working with Laury Fincham of KEF, he did some research on the audibility of the phase effects of the just-developed LR. In fact, he used the same technique as Malcolm did - running the signal through an allpass. Siegfried at that time was able to hear the effects reliably with some special test signals but not with music.
Isn't audio wonderful??
jan didden
How do you figure? It's not known from the start which files have been processed and which files have not. So how can there be any reference?
There should be a known, non-processed reference file so that it can be A/B'd against the six unknown files.
Sure, you can pick one of the unknown files and A/B it with the remainder of the files, and then try and sort out which are which but that's a rather messy approach in my opinion.
se
That's if you care about the order and which group is which. If you don't, then you can eliminate the degeneracies. For example, let's consider two groups, A and B, with six objects to be divided into those two groups. Number the objects from 1 to 6. Call group A the one in which we place the first object.
So group A can consist of 123, 124, 125, 126, 134, 135, 136, 145, 146, and 156. Other combinations are degenerate.
So group A can consist of 123, 124, 125, 126, 134, 135, 136, 145, 146, and 156. Other combinations are degenerate.
That's the approach I took and what seemed the simplest to me. I just compared all files to the first one. Same? Group A. Different? Group B. I didn't bother to compare the other files to each other, only to the first one since I knew going in there would be only 2 groups.
Funny, one of my comments to Jan was that they sounded like they had been cut up into bands and put back together, maybe thru an encode, decode scheme. I was partly right, but only partly.
Funny, one of my comments to Jan was that they sounded like they had been cut up into bands and put back together, maybe thru an encode, decode scheme. I was partly right, but only partly.
- Status
- Not open for further replies.