|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
65.19.76.104
In Reply to: RE: Enough Computer Audio Mythology posted by SBGK on December 12, 2014 at 04:35:35
There are good reasons why changing the control program might affect the sound quality. It might not be necessary for the control program to be running to affect the sound on playback. It might not be necessary for the control program to even be loaded in memory to have an effect. For example, two different control programs could result in the play program or the audio samples being loaded in different places in memory. Also, they could have left the CPU caches in different states.
If you want to have a fair shot at a valid experiment you would need to verify that both control programs left RAM memory in the exact same state, which means that the audio data has not only to be the same, it has to be stored in the same locations and every other memory location in the RAM has to be identical. In other words, the entire RAM has to be scrubbed except for the audio. At the least this will require verification that the audio data is loaded in the same locations and is unchanged. Scrubbing the rest of the RAM can probably be accomplished by rebooting the computer, if the audio data in RAM can be somehow protected. (I say "probably" because there still could be logic state that does not get reset at the time of a software reboot.)Beware of the Turing tar-pit in which everything is possible but nothing of interest is easy.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Follow Ups:
"There are good reasons why changing the control program might affect the sound quality."
Or are there not good reasons? What exactly are you saying? If you are trying to assert that a "control program might affect the sound quality" you certainly have a roundabout and equivocal way of saying so.
C'mon, stake out a position!
People here have scolded me for asking to see measurements to back up audible claims by saying that hearing a difference lead to subsequent changes in measurements. The implication of course (if indeed not the assertion) is that hearing is superior to measurement.
Any guess as to when we'll see measurements to back up or refute any claims that "control programs might affect the sound quality?"
JE
I already staked out my position. I would assume that the readers are intelligent enough to understand what I was saying. This is a computer audio forum and I would expect people who say that something is possible or impossible to have some experience with computer programs and computer programming. I would expect them to understand that buggy programs can do things that astonish their creators.
It is hard to communicate effectively with fools and trolls.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
my mqnplayer is there to try with different control programs, no one that has tried it says they haven't heard a difference. Maybe self selecting, but without suitable measurements it's all we've got.
I'm going to create a program that reads and outputs a wav file to see if the effect can be heard in different players, the sound can be made better or worse.
talk of buggy software is just childish, the last 2 versions only changed in terms of process/thread priority and switching thread boost off and the code is at such a minimum that it just reads file to ram and writes from ram to ram in play program, the only differences are settings in opening the file, ram alignment, ram security settings and compile/linker settings. The play program is even more optimised being just 3.2 kb as until recently I thought the control optimisation didn't matter.
If the effect is repeatable and other people are reporting the effect on their systems without prompting then don't see the requirement to ensure ram is identical etc, if others want to test that then am happy to provide code to help them do that. Sometimes an effect is too much for some people and they prefer one version over another. Needless to say most people trying it have better resolving systems than I do. They seem to like the bugs.
http://mqnplayer.blogspot.co.uk/
People have reported hearing differences that are "impossible", where I have put quotes to indicate that perhaps these people don't know what they are talking about. I assume if many people report hearing "impossible" differences that at least some of them did hear real differences. I am not a BS "objectivist" who claims that this must be impossible and therefore people making these reports must be liars or fools.
My objection is not with the people making listening reports. My objection is with the people claiming that what they heard was "impossible". From all I can tell, the people claiming that theses differences are "impossible" have little or no clue regarding the inner workings of computer hardware and software as relates to audio playback. This is easily deduced from reading their posts.
I have yet to see a computer program that worked perfectly, except those that I wrote myself, of course. And in one case, I went so far as to make a bet at 10,000 to 1 odds there were no bugs, but nobody took me up on the bet. :-) They would have lost. See link.
Lots of audiophiles prefer distorted playback. It may be because their goal is just to hear what they like without reference to what was actually recorded, possibly because they have no clue as to how the recording was supposed to sound. This is the usual situation for studio recordings that are highly processed. For live recordings of acoustic music, music lovers with frequent concert experiences have a much better chance of a reference, but even here it can be dodgy. IMO, the only way to get a solid handle on this issue is to make and playback recordings of live performances. If one is focusing on digital distortions, one can also do tape - amp - speaker vs. tape - digital; digital - amp -speaker tests and use the tape playback as the reference. However, if differences are found there won't be a way of distinguishing whether the differences come from the ADC, the file format, or the DAC. More complex tests will be needed to isolate the cause(s) of differences, and these won't be easy or definitive.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
if it's a bug then it would affect the sound, the player has been tested and is bit perfect.think the people who try it know more than I do about computer audio, so trust their feedback. Some like a brighter more analytical sound than I do.
the effect seems to be able to be stored in the bits in ram, that's the new thing to me, which I wouldn't have believed until I heard it, because I've compared linear powered rips with normal rips before and thought they sounded different, but thought it was impossible. This seems to be the same effect.
Whether people no longer need to worry about linear ps when ripping if they can just load the file in an optimal way remains to be seen, I've never been that careful when ripping.
http://mqnplayer.blogspot.co.uk/
Edits: 12/15/14
Each bit in DRAM is stored in a tiny capacitor. This capacitor is charged with a small number of electrons. Below a certain number it's a 0. Above that number a 1. There is a dedicated capacitor for each bit and a transistor that serves as a switch to connect this bit onto a sense wire, where the number of electrons can be measured. Now the problem is that the electrons slowly leak away for various physical reasons. (This depends on the temperature of the substrate, residual radioactivity, etc...) If there were initially a large number of electrons that represented a 1 then after leakage they may enter in to the ?? range, or even leak down to look like a 0. To deal with this problem, there is a memory controller. It reads every memory location (refresh operation) about once a second. Each bit that is read is rewritten back so that it is close to the ideal number of electrons for a 0 or a 1. This is the equivalent to looking at a smudged image by eye and recopying on a clean piece of paper.
It seems most unlikely that hidden information will survive after a few seconds of recopying. It seems much more likely that the data in RAM is not identical in the two test cases, or that there is data stored elsewhere in the computer system that is different. It is possible to compare the contents of RAM memory by loading special software with a special boot loader program, but probably not all other parts of system state. I would expect things like processor cache may differ and this may have important implications on system performance. Unfortunately, these factors depend on the specific processor and motherboard models.
There are many ways to explore this rabbit hole in depth. However, IMO unless the entire contents of RAM is validated to be identical in the two cases there is little point in proceeding further. I recall the WAV to FLAC to WAV vs. original WAV issue. At first glance, the audio samples in the two WAV files were identical, but after investigating further, the two files were not identical, as their meta-data differed. I see no reason to believe this case is any different.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
how could changing the priority of the process affect the bits in RAM ? There's something about how the data is stored that's affecting the sound, all I'm doing is reading it in and writing it out, there's no data manipulation. Read in 4k at a time. Some people say the noise in digital audio is due to rf noise, is it possible that cpu rf noise is getting modulated into the bits ? others say digital audio should be thought of as analogue, in which case a modulated signal would be possible.
http://mqnplayer.blogspot.co.uk/
None of this noise "should" make it through the DAC. However, it does. How and why will depend on the specific equipment. Some people have looked at the noise spectrum of most DACs and there is a lot of RF noise that doesn't show up in typical spec sheet or magazine review measurements. Miksa (posts on computer audiophiles) has done a lot of these.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
The forum has been relatively peaceful the last several months.
"The forum has been relatively peaceful the last several months."
This forum. The action is elsewhere...
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: