Trench War.
Sept 20, 2011 10:58:33 GMT
Post by lark on Sept 20, 2011 10:58:33 GMT
Sure.
What is being said is that the improving of power supplies is having an audible effect on rips.
Agreed.
We are assuming, yes assuming, that this is due to the removal of a certain amount of noise, mostly from the computers own power supplies.
This assumption is where things start to go awry. Power supplies create noise in the analogue domain in which the digital data is being transferred over but that noise can never make it into the digital data (that would change the data and would be detected by checksums (used throughout the digital process) and retransmitted until it is perfect (ie - no errors)). Any other artefacts present in the analogue signal are discarded only only the digital data remains - hence no noise.
This is a common and well known improvement in the analogue domain and seems to be beneficial in the digital one too.
No. It is applicable in the analogue domain, but doesn't even make sense in the digital domain.
It is a finding that does not come with a 100% definite explanation.
Agreed - there is no explanation to describe the effects we're discussing because the digital systems guarantee that it can not occur.
You do realise the for someone to upload and then download a file from filemail.com that data will hop through at least 24 machines, servers, routers and switches in each direction that can and will do what ever they want with the data as it goes through. This happens twice, once for upload and once for download...
Of course! If two files go through the same procedure, via the same route/machines etc. then they will be effected equally. Thus differences are still perceivable at the recipients end even though both files may have been degraded. This also points to the significance of the finding that it is discernible after all this travelling by the files.
My point here is more that there is sooooo much going on here - that's all out of our control - that if one believes that such power supply tweaks can improve (I'll get back to 'improve' in a bit) the 'quality' of the data transmitted between a CD Rom reader and the storage device, then one should be horrified about the effects of transmitting a file over the internet, through thousands of KM of copper, fibre, through many machines, routers, switches etc. - But there is nothing to fear, this process will not modify the data at all.
Add to this that a received file can sound superior to one I have ripped locally and the mind boggles.
I completely understand that this is mind boggling - that's because it's impossible.
Your Wiki link on checksums never actually states that the checksum procedure is totally 100% perfect.
checksums as implied by its name checksums are simple arithmetic - really. Digital data is numbers (represented in binary) but it is just a set of numbers - computers ONLY deal with numbers they know nothing else, it's just that the numbering system used is binary as opposed to the better known decimal system. But it is all just numbers. Digital images are just numbers, digital video is just numbers, digital audio is just numbers, a word document is just numbers - all stored in the binary numbering system.
So to give an example of how a checksum works (and why they are 100%). If we had the data file that contained the numbers:
1, 2, 3, 4 and 5
This would be stored in binary as:
00000001, 00000010, 00000011, 00000100 and 00000101
Now a really bad but simple checksum could be to add the numbers together:
1+2+3+4+5 = 15
So, the checksum algorithm is known on both ends - this needs to be the case so that it can be generated on one end and checked on the other. So our super simple but dumb and bad checksum algorithm is simply the sum of the data (15). Now if I transmit that data over some noisy analogue medium I can send it and the checksum to verify it was received correctly. So our protocol will be "Transmit 5 numbers (or 5 * 8 bits) followed by another 8 bits for the checksum"
So we send the data, and the receiver gets 8,2,3,4,5,15 - this is obviously wrong since we add up 8+2+3+4+5 and get 22, we then compare that to the next number in the packet and see that 15 != 22 - corruption occurred - retransmission required. This process will be repeated until a predetermined number of unsuccessful attempts are reached or the checksum matches.
Now, this example is bad as it doesn't find errors well - actual checksum algorithms are still just arithmetic on a set of numbers. For most checksum algorithms a single bit error will yield a vastly different checksum result.
So by the shear nature that data is numbers, checksums are arithmetic on these numbers - they can only fail if maths is broken - I seriously doubt that maths is broken.
It only goes as far to say that equal checksums indicate that
"If the checksums match, the data were almost certainly not altered (either intentionally or unintentionally)."
Sorry, yes that is not stated, but it's somewhat given by the nature of what they are doing - see my explanation above.
The ONLY sites I could find that did claim perfection were the producers of the checksum programmes!
I think they are referring to the perfection of their implementation of the checksum algorithms, not the algorithms them selves - these are proved my the (seriously freaky) mathematicians that created the algorithms.
So equally, from my stand-point, there is no proof that the checksum is infallibly perfect or accurate enough
If my explanation doesn't satisfy you, do some more research and you'll find that they are 100% accurate.
Those who argue that there is a sound difference are doing so from unexplainable waters, we cannot MEASURE why this is happening with the knowledge and tools available.
No analogue measurements are required - we can verify the source data instead (easier and more accurate).
No they are not! They are set of on-and-off states, like any other digital file (binary code). On and off comes from, er, oh yeah, power supplies!
I'm not sure the sarcasm is really required. But, no - digital data is numbers represented as on off states stored a binary number as described above. The on off states are not really produced by a power supply like you're suggesting.
the digital domain... was created by humans, to work in a specific way, which it does every time - there is nothing to 'work out' as we humans designed it.
Us humans have designed a lot of things to work in a "specific way" and they fail to do so ! Are you claiming this human creation is the ONLY perfect one?
If you read my comment again, you'll notice that I was referring to the fact that we don't need to 'work anything out' as we designed it, as opposed to the analogue world which we have to interpret (we didn't design it, we don't really know how it works) Digital though was designed and is 100% understood because (no unknowns). So my comment has nothing to do with being 'perfect' it as to to with what we know about it!
In the meantime you can be happy with your perfect digital files while others can be happier with their better ones
Ok, back to the 'perfect', 'better' data scenario. Data is just numbers - unlike an analogue wave form, there is no almost perfect, partially perfect, or similar to - it is always exactly the same set of numbers (guaranteed by checksums etc), always.
So if a music file consisted of the numbers 99, 12, 323212, 312355, 1233234, 132322 that represented the waveform of the analogue signal captured by the ADC then how do you suggest that these numbers can be improved by using a super duper power supply? a better 99? a smoother 12? a deeper 323212? - if these numbers change during transmission, they will be rejected by the receiver and a retransmission occurs until it's correct. That data MUST get through 100% intact and 100% the same or the rip failed and your software will tell you such - this can happen with faulty hardware, but is very rare. But either way, you'll be told - rip bad!
A rip does nothing more that take a copy of some numbers on a spinning plastic disc and copies the numbers to a different media - You could even use a microscope and pen and paper and transfer the data to a sheet of paper by hand if you wanted to - it would still be 100% the same quality when you then retyped those numbers back into a computer. Regardless of the quality of your heart or pacemaker..