|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
206.255.203.68
In Reply to: Could this be the answer to the objectivist/subjectivist debate? posted by The Real Dick Hertz on February 27, 2007 at 13:12:18:
Will a second generation recording of a system in use be good enough to resolve the subtle differences involved? So you bring your $200k worth of recording equipment to your listening room and make a baseline recording. Then change something and make a new recording. Then perform the analysis between the two signals.Sounds great on paper, but impractical and fraught with unvalidated assumptions.
Follow Ups:
You are saying that an audible difference in signals might yield NO difference between recordings of that signal and the unchanged signal?In other words, that whatever has changed will be *totally invisible* to the less than $200K recording process and will cause no difference at ALL between the tracks made with it (kind of like the way vampires don't cast reflections in mirrors)?? Are there signal characteristics that get removed (not just get distorted or corrupted or mixed with added noise -- but vanish *completely*) in passing through gear that doesn't cost enough $$?
I can think of two cases where that could possibly be true:
...Signals outside the bandwidth of the lower$$ gear-- but soundcards that record at 192kHz have bandwidth as high as any used in studios, and can be had for under $150. So that can be easily avoided.
...Or signals so low in level that they are below the noise floor of the sub$$ gear -- but do all the audio benefits of tweaks and cables only occur at -105dB below the peak playback levels?
..Maybe there are other classes of stealthy sounds in one recording that can leave no trail in another recording, who knows for certain? Pretty darn unlikely, though, I'd say.The "not good enough equipment" argument maybe holds when talking about degrading a signal, garbage combining with other garbage sounding like the same garbage. But the test mentioned detects whether any change has happened at all.
Are you going to multi-mike the second generation recording in a user's listening room? Use a minimal mike approach? Will that be the same approach as was taken with the first generation of the recording? Is that going to retain all the original recording's perspective? And S/N ratio? If I understand the approach correctly, we are re-recording a recording in another room. Twice.
EStat,"Is that going to retain all the original recording's perspective?"
Sorry, I didn't follow that the first time I read it, and need to respond.
I need to get it clear that IT DOESN'T MATTER if the perspective is the same, or if the recorder's response isn't dead flat or if the S/N is degraded some, etc.. Diffmaker is looking for differences, only. As long as the recording situation is the same for both recordings. Things like bad S/N will make a difference in the two recordings (noise isn't the same every time and doesn't subtract out). All non-silly errors I can think of that are makable by the program will be in the direction of showing a (possibly nonreal) difference, rather than showing "no difference" (that takes an *awful* lot of things going right to achieve).
From the other post:I guess I was misled by the block diagram whereby you first "make a Reference recording". What what you've said, you capture line or speaker signals which are not really "recordings" in the usual sense.
I now notice that this is your product, so naturally you have devoted far more time to understand its principles! I genuinely wish you good luck in getting real audio designers (as opposed to the hobbyists here) to use the tool.
I suspect you haven't read the documentation that comes with the program. No offense, but there's a lot written there about that. Here's a summary:Recording with mics in a normal listening room isn't likely to be usable, the thing is insanely sensitive to even teeny differences. Things like the mic or speaker moved several mils, very exact positions of reflective sources including the person doing the recording in the room, even in other rooms that open into the measurement room, maybe air currents and temperature changes). I've not obtained a good silent difference track with acoustical signals when testing with no changes at all between recordings (a good verification test for setups).
For similar reasons, I doubt it would work with turntable sources either (though I haven't tried) because of changes in the vinyl after the first playing. Maybe if it were done on different days after the plastic has relaxed again it might work, if the turntable speed was consistent enough.
But electrical signals can be recorded from speaker terminals or at line levels between any components in the chain. As long as what is being changed comes before the place where you record from, you should be able to pick up differences in the audio electrical signal.
So devices to test might include cables, CD treatments, contact enhancers, AC Power enhancers, vibration control devices, etc. (unless they are having their effect on the listener by some way other than through the electrical audio signal).
So things like speaker drivers (which are generally agreed to make a difference anyway), room resonators, other stuff that doesn't act on the electrical signal probably can't be tested with Diffmaker, at least not without special isolated conditions and extreme care. It doesn't do everything, but neither does any other test. Use where appropriate.
I'm really not out to prove that things don't make a difference (that couldn't be proven anyway unless every possible situation was evaluated). But the reverse is provable, and by only ONE universally repeatable example! I keep seeing finger-wagging audiophile posts here and elsewhere about how engineers aren't measuring the right thing. So here's an attempt to do that. I can't do scientific investigation and find those measurements without actual evidence (as opposed to testimonials), so the idea was to get evidence where it may be gettable, and also to maybe give me and others a feel for the degree of difference being made (or not made).
I'm not trying to take away anyone's toys, OK?
It will take only one repeatable example of an audible effect (that isn't explained by current theory) to allow actual research on it to start (and fame for the lucky engineer who gets first crack at it, too). Help me out here, if you want to (not just EStat, anyone else, too). If you really want someone to "measure the right things", that is.
...you then have my blessing. Mess with my toys however... ;-)Seriously though, sounds like you've created a valuable tool.
x
This post is made possible by the generous support of people like you and our sponsors: