Fidelizer Pro - Real or Snake Oil?
Status
Not open for further replies.
Feb 19, 2016 at 1:48 PM Post #271 of 683
So here's your cue @WindowsX--KeithEmo has pointed out that a PC will generate significant amounts of jitter on its S/PDIF coax or optical outputs, and these can be read directly (as opposed to USB audio)--why not directly measure jitter on these with Fidelizer on and off instead?

 
It's possible with audio interface having digital input/output to perform closed loop recording. I'm not sure if that is possible, maybe we'll need two audio interfaces? I'm not sure.
 
Anyway, it's not like I have such devices at hand to perform this nor any device will work. I used to have RME Fireface 400 but it's broken not so I can't test it myself anymore. Maybe someone can take it from here following my examples for real world data. But I doubt you'll find something audible with analog metrics.
 
  But you didn't measure anything audible, and besides which, as soon as it has to be time-aligned, it's no longer completely bit perfect.

 
It'd be crazy if you can measure anything audible in pure software environment. It's all in virtual sound card. You should be surprised that I can even measure something out there to show you here.
 
By the way, how can you measure different recording files without time alignment? Aligning doesn't break bit-perfect because audio stream aren't altered in anyway except choosing where of stream to compare. There might be some rounding error to consider like I said before but all measurements showed 0 dB so it's all good.
 
 
 
Well, it seems to be only me who took actions to find out. I hope someone will run the similar experiments to share with us soon.
 
Regards,
Windows X
 
Feb 19, 2016 at 1:54 PM Post #272 of 683
It is your job to get your believers to measure and prove an audible difference.

And if you yourself feel there'll be no audible difference in the analogue output, then I don't see why you claim what you do.
 
Feb 19, 2016 at 2:23 PM Post #273 of 683
It is your job to get your believers to measure and prove an audible difference.

And if you yourself feel there'll be no audible difference in the analogue output, then I don't see why you claim what you do.

 
Your audible threshold will make no different between any digital transport on DAC with (random)jitter immuned design.
 
Regards,
Windows X
 
Feb 19, 2016 at 3:21 PM Post #274 of 683
It is this jitter, generated by the computer, that Fidelizer claims to reduce.

 
Did Fidelizer actually claim to reduce jitter?
 
I don't recall seeing that.
 
Feb 19, 2016 at 3:31 PM Post #275 of 683
   
By the way, how can you measure different recording files without time alignment? Aligning doesn't break bit-perfect because audio stream aren't altered in anyway except choosing where of stream to compare. There might be some rounding error to consider like I said before but all measurements showed 0 dB so it's all good.
 

You can't. If it's time-aligning by a non integer number of samples though, it's no longer bit perfect. There are several other possibilities too - honestly, I don't know for sure (nor do I really care) why you are only getting a diff down to -150 dB rather than -300, like you did with two copies of the same file. At the end of the day, -150dB is still fantastically good, and there's no way you could even come close to hearing errors at that level. Hell, some of the best analog equipment in the world can't get a broadband noise floor that low.
 
Feb 19, 2016 at 3:34 PM Post #276 of 683
   
Are you sure you understood X's data?
 
Yes, perhaps it did have an effect (although it may have also been quantization), but the results in this specific scenario are in the range of -130 dB, which is inaudible.

 
I'm not making any claims either way.  I do know that 1 test over a short time with nothing else really going on in the pc isn't representative of reality.
 
I've been a software engineer for over 20 years, and I know a bit about how cpus work.  Windows could be introducing mutexes and semaphores across broad swathes of it's internal sound system for reasons we have no idea about.  Basically, the windows kernel is a black box and none of us can predict what can happen in all curcumstances.  Having a momentary spike in CPU usage may cause cache wipes, and require data to be retrieved from main memory which, compared to L1, L2, and L3 cache, take an eternity.  It's entirely feasible that altering settings in the audio engine in windows could mitigate these things to some degree.  Or not.  I won't pretend to know, I haven't tried the product.
 
I do know that often network activity, especially wireless traffic, can introduce tics into the sound I hear.  I suspect this is because of system interrupts throwing timings off in other subsystems.
 
More data would need to be collected over say, a week or two.  If the Fidalizer software could detect and log instances where it is actively helping, that would be a good thing to see.
 
Feb 19, 2016 at 3:46 PM Post #277 of 683
  You can't. If it's time-aligning by a non integer number of samples though, it's no longer bit perfect. There are several other possibilities too - honestly, I don't know for sure (nor do I really care) why you are only getting a diff down to -150 dB rather than -300, like you did with two copies of the same file. At the end of the day, -150dB is still fantastically good, and there's no way you could even come close to hearing errors at that level. Hell, some of the best analog equipment in the world can't get a broadband noise floor that low.

 
As I said before, 300dB for identical files and about 150dB for silence added files. Did you read my posts in few pages back? And the proof of concept in this experiment is to find measurable changes of bit-perfect application working in pure software environment whether Fidelizer is snake oil that does nothing to contribute audio for the better or not. And the result is it's not snake oil and I have concrete data to back this up. It's up to individuals to decide if inaudible data will signify anything.
 
   
I'm not making any claims either way.  I do know that 1 test over a short time with nothing else really going on in the pc isn't representative of reality.
 
I've been a software engineer for over 20 years, and I know a bit about how cpus work.  Windows could be introducing mutexes and semaphores across broad swathes of it's internal sound system for reasons we have no idea about.  Basically, the windows kernel is a black box and none of us can predict what can happen in all curcumstances.  Having a momentary spike in CPU usage may cause cache wipes, and require data to be retrieved from main memory which, compared to L1, L2, and L3 cache, take an eternity.  It's entirely feasible that altering settings in the audio engine in windows could mitigate these things to some degree.  Or not.  I won't pretend to know, I haven't tried the product.
 
I do know that often network activity, especially wireless traffic, can introduce tics into the sound I hear.  I suspect this is because of system interrupts throwing timings off in other subsystems.
 
More data would need to be collected over say, a week or two.  If the Fidalizer software could detect and log instances where it is actively helping, that would be a good thing to see.

 
I'm also software engineer, writing Fidelizer with understanding of how computer and OS works. You may read about what Fidelizer does in details from here. I wrote it years ago so please forgive my crude writing. :)
 
http://www.fidelizer-audio.com/about-fidelizer/
 
I think it's possible to log processes to monitor interrupts and resource scheduling. However, it's still questionable to interpret that kind of data for audible changes. It seems most people won't take anything outside RMAA standards as proof. But I'm satisfied now with data I gathered.
 
Regards,
Windows X
 
Feb 19, 2016 at 3:57 PM Post #278 of 683
The .Net framework exposes a lot of the instrumentation that windows collects.  I'm sure you can also get to in using c++ as well.  
 
Just don't let the closed minded "Open Minded Scientists" here get you down.  I live by the idea that we don't know what we don't know, and that future discoveries can change our entire understanding of fundamental things.  It's arrogant to assume that you know all that can be known about almost any subject.  I'm not sure many of the regulars here fell the same way. :)
 
Feb 19, 2016 at 4:13 PM Post #279 of 683
You could do that - sort of..... except.....
 
1)
 
It's really hard to measure jitter directly - since we're talking about variations in a clock frequency on the order of single-digit nanoseconds or less. This is something that requires specialized equipment to measure, and absolutely cannot be measured with a simple oscilloscope or meter. The normal way of detecting jitter is to look for the distortion sidebands that result from jitter being present when you send a digital audio signal through a DAC that is sensitive to it. (In other words, it's easy to measure how jitter affects a particular DAC, but quite difficult to measure jitter directly. The good news there is that the same measurements work for any type of input.)
 
2)
 
Assuming that Fidelizer functionally does what it claims, and turns off several unnecessary processes running in Windows, then it almost certainly will reduce jitter - on some computers - to some degree. However, confirming that won't tell you whether the improvement it makes is audible or not. (At best, it would tell you if there was no difference - which is actually pretty unlikely.)
 
The real, problem, however, is that the amount of improvement that Fidelizer makes will depend largely on the computer you try it with. It will probably make at least some difference on most computers, but how much difference will depend on the specific computer you try it with. And then, once you determine that it reduces the jitter on the computer's output, how much difference THAT will make, and whether it will be audible or not, will depend to a major extent on the DAC you try it with. Some DACs are very sensitive to jitter, while others are almost totally immune to it.... This means that your test is very unlikely to be a simple "yes it works or no it doesn't" and much more likely to be "does it work, and, if so, how well, with this or that particular DAC, when connected to this or that particular computer".
 
However, in the simplest case, measuring it on a "typical" computer, with a DAC that lacks any sort or ASRC or other jitter reduction mechanism, connected via a Coax connection, and playing a few simple test files and looking at the spectrum of the audio output, should be able to give you an idea of whether it actually does ANYTHING or not.
 
Quote:
So here's your cue @WindowsX--KeithEmo has pointed out that a PC will generate significant amounts of jitter on its S/PDIF coax or optical outputs, and these can be read directly (as opposed to USB audio)--why not directly measure jitter on these with Fidelizer on and off instead?

 
Feb 19, 2016 at 4:39 PM Post #280 of 683
The author is a programmer, and so looks at things from a programming perspective..... whereas I am more inclined to look at them from a hardware engineering perspective.
 
There are only two "components" of data; the data itself and the timing with which it is delivered. Since we all seem to agree that the numbers themselves aren't being altered, the only remaining area which can be affected is the timing... and it has been a widely accepted fact for along time that PCs don't deliver very consistent clock timing. (The author looks at it as eliminating or adjusting processes so as to deliver the data on a more accurate time schedule; I tend to look at it as the time schedule being important, and improving its accuracy being likely to produce some sort of improvement, even though I may not be especially interested in the details of which processes are eliminated and which ones are adjusted.)
 
The term jitter can be to apply to timing errors in general, or to specific types of timing errors... and I'm using it in the general collective sense of "any errors in data timing". Likewise, whatever term you choose to apply, it can be easily demonstrated that at least some DACs produce significant and audible errors in their analog audio response when presented with data that has timing errors. (This is well understood; for a given known timing error, and a known signal, you can calculate the precise amount and type of distortion that will result with a given type of DAC. And, for a given "theoretically perfect" DAC, you can calculate the amount of timing error it was subjected to based on an analysis of the distortion spectra.)
 
The author of Fidelizer seems to be claiming simply that his program optimizes several areas of Windows performance in ways that seem to be consonant with better audio performance. That it improves audio quality by reducing timing variations and errors (which I would term as jitter), and also that it can reduce or eliminate some dropouts that result from certain "non-optimal processing situations" (my phrase - not his) - both of which seem like sensible claims for a program that optimizes timing and eliminates unnecessary processes. To me, the jitter aspect should be relatively easy to measure and confirm; I would assume that a reduction in dropouts could be confirmed by statistical analysis...
 
Quote:
   
Did Fidelizer actually claim to reduce jitter?
 
I don't recall seeing that.

 
Feb 19, 2016 at 4:45 PM Post #281 of 683
  The .Net framework exposes a lot of the instrumentation that windows collects.  I'm sure you can also get to in using c++ as well.  
 
Just don't let the closed minded "Open Minded Scientists" here get you down.  I live by the idea that we don't know what we don't know, and that future discoveries can change our entire understanding of fundamental things.  It's arrogant to assume that you know all that can be known about almost any subject.  I'm not sure many of the regulars here fell the same way. :)

wink_face.gif

 
most guys here are not closed minded, they're just skeptical and tend to simply reject things that aren't properly demonstrated.
and within those guys, a few don't express it with the kindest posts. I don't feel like the most passionate posts will ever represent the overall vision of any community.
 
Feb 19, 2016 at 5:56 PM Post #282 of 683
Why even bother participating in the HF Sound Science forum?  I'd just provide a link to the 6moons review and laugh all the way to the bank.  Big spending audiophiles don't need anything more to make up their minds.
 
http://www.6moons.com/audioreviews2/fidelizer/4.html
While music sounded more transparent, tonally correct and dimensional than not using Fidelizer at all, the Pro version was 
significantly
 superior in every way, particularly in overall transparency and in its rendering of space and layering.

 
I was simply curious to understand more about what is supposedly making this "significantly superior" audio quality that reviewers and customers believe they are hearing.  I couldn't hear anything different.  I still don't understand what is possibly making this kind of improvement, if at all, or my definition of significant is lacking.  
 
Feb 19, 2016 at 6:28 PM Post #283 of 683
  Why even bother participating in the HF Sound Science forum?  I'd just provide a link to the 6moons review and laugh all the way to the bank.  Big spending audiophiles don't need anything more to make up their minds.
 
http://www.6moons.com/audioreviews2/fidelizer/4.html
 
I was simply curious to understand more about what is supposedly making this "significantly superior" audio quality that reviewers and customers believe they are hearing.  I couldn't hear anything different.  I still don't understand what is possibly making this kind of improvement, if at all, or my definition of significant is lacking.  

 
After losing words to express for my data, you resorted to quote some audiophile reviews just to spout this nonsense? How stupid...
 
I'm grateful to see some opened mind people who know about stuff they're discussing about but lately there's too many of ungrateful people who keep degenerating the advancement of audio measurements. I hope to find some better place to discuss about audio science without prejudge some days.
 
Regards,
Windows X
 
Feb 19, 2016 at 7:04 PM Post #284 of 683
   
After losing words to express for my data, you resorted to quote some audiophile reviews just to spout this nonsense? How stupid...
 
I'm grateful to see some opened mind people who know about stuff they're discussing about but lately there's too many of ungrateful people who keep degenerating the advancement of audio measurements. I hope to find some better place to discuss about audio science without prejudge some days.
 
Regards,
Windows X

 
I posted the link to Achimago's blog that had some real audio measurements that made sense to me, and I have been reading the comments in this thread looking for any information to support or refute the data from the blog.  Other than some discomfort from a few individuals in an audio asylum discussion with the frequencies used for the IMD test, I really haven't seen anything to make me discount the testing of this product.  
 
Am I "degenerating" (denigrating maybe?) the advancement of audio measurements?  I don't understand what you mean.
 
 I apologize to you for my preconceptions, but I see your product in the same way I see those banner ads in websites that claim their software can make your PC run faster.  
 
Feb 19, 2016 at 7:16 PM Post #285 of 683
   
I posted the link to Achimago's blog that had some real audio measurements that made sense to me, and I have been reading the comments in this thread looking for any information to support or refute the data from the blog.  Other than some discomfort from a few individuals in an audio asylum discussion with the frequencies used for the IMD test, I really haven't seen anything to make me discount the testing of this product.  
 
Am I "degenerating" (denigrating maybe?) the advancement of audio measurements?  I don't understand what you mean.
 
 I apologize to you for my preconceptions, but I see your product in the same way I see those banner ads in websites that claim their software can make your PC run faster.  

 
I'm the first one who posted Archimago's link in first page. To begin with, archimago's measurements are for measuring AD/DA performance with analog metrics. That kind of method can't measure anything directly in digital domain.
 
Your act to quote outside messages from reviewing website just to attack products is people are degenerating. It's unproductive and uncivilized. I never referenced reviewing website and tried to talk about measurements, method and data here.
 
Since only basic audio measurements can make sense to you, I suggest you to take some pro audio course or some digital audio classes. Or you can just try Fidelizer and see for yourself if it works for you or not.
 
Regards,
Windows X
 
Status
Not open for further replies.

Users who are viewing this thread

Back
Top