Home Propeller Head Plaza

Technical and scientific discussion of amps, cables and other topics.

Re: Even if the software were perfect ....

"nothing that you can do will take the sampling uncertainty out of the equation."

No, but you can design with the intent of reducing audible distortions in the final product by addressing specificly noted shortcomings of past effeorts with state of the art technology if necessary until the specific complained about effect gets minimized to the limits of technology or to the point that the problems are effectively eliminated.

"providing a square wave or impulse will not fix the problem "

my bad I only recomended squares and impulses since I believe that communications theory says we will be able to get more accurate transfer function approximations from the device under test using these types of stimuli than we would using some random audio sample. I didn't mean to imply that I KNOW that these particular waveforms are any more useful in guaranteeing perfect time alignment of the sample sets being compared, other than what may be implied by the fact that a very strong reference quality signal generator market exists with a wide range of highly accurate commercial products being widely available which could potentially be leveraged if it were deemed to be part of a solution which worked better to achieve the desired results. On the contrary....

"I am assuming that Mr. Waslo did a good job on the auto-alignment portion of his differencing software,"

With all due respect, I gottta say that's brave. I'm proposing here and now that doing this correctly may even be what lies at the root of why previous attempts at this particular type of analysis may not have yeilded the desired results. I'm not an expert on how to or the motivation of which spot on a signal to choose which to trigger but I'm starting to think triggering on the signal properly and the assumptions made deciding just where to trigge may be critical to getting good results. I believe this is quite a bit more precise and critical than the simpler time aligniong software (cakewalk etc.) out there used to line up music tracks for musician song editing would allow for. For example: I'm guessing the nuances of time and frequency information differences between cables under test might be quite a bit more subtle than the time alignment and signal stretching effects necessary to satissfy a drummer that his drum beat track is sufficiently lined up with his mate guitar riff in order to sufficiently "rock on". In other words we may require much more accurate time alignment software/hardware capabilities such as on the order of what digital sampling scopes are capable of in their triggerring algoriyhms and circuitry rather than what a musician software/hardware package allows for. It would be interesting to know how this software stacks up and what assumptions ar being made. Not that I am enough of a signel processing whiz to be useful at analyzing the algorithms myself without more research.

"Nothing can be done either about the inherent and intrisnic amount of jitter the soundcard and computer system have, a reference sq wave or an impulse would not provide any better performance here."

If you are referring to the synching of the A to D and D to A operations in time with the signal being sampled as well as to each other, well I have a couple things to say about that. First, If the errors of each domain conversion are reduced to below the level of what is audible then it is my belief that this is not necessary in the first place as long as proper time alignment algorithms are used for doing the difference comparison. Perhaps this downfall of modern equipment you are seeing is related to this application pushing the limits of what the hardware is intended to do, ie the combination of error due to digitizing and due to re generation of the analog signal. Perhaps with this in mind, if it indeed turns out to be part of the problem the next generation of or special niche of soundcards will be designed to address this issue.

If in fact you are referring to the variation of clock to clock edge timing accuracy in an already synched system, well, that just speaks to some unfilled sound card niche since these soundcard manufacturers of these "pro" caliber soundcards you mention later in your post need to quit cheaping out. That of course is assuming you have heard some digital technology which is jitter free enough that these dream spec sound cards we are going to dream up might aspire to have if the soundcard manufacturers creating a new card were to try to cater to the desires of a consumer in this potential market.

"To this day, WE DO NOT HAVE A COMPREHENSIVE SET OF MEASUREMENTS THAT WILL ALLOW US TO FULLY DETERMINE WHAT THE LEVEL OF SONIC ACCURACY TO THE HUMAN EAR WILL BE PERCEIVED AS." and "

admittedly my statement regarding technology being able to address all humans hearing capabilities is more or less unproveable. Sorry I take it back. I still believe the point I was trying to make is valid though that if you weren't having success identifying clearly audible differences in a black box under test using equipment which is good enough perhaps it is the assumptions made during the data analysis that is flawed not the equipment. This brings me back to the point I raised above about whether best practice regarding signal triggering and the assumptions used to time align the before and after comparisons are being used to gurantee accuracy of the desired result.



This post is made possible by the generous support of people like you and our sponsors:
  VH Audio  


Follow Ups Full Thread
Follow Ups
  • Re: Even if the software were perfect .... - Ugly 14:51:45 03/03/07 (0)


You can not post to an archived thread.