Linear Powered Rips & flash drives sound better - Alex was right !

[alfe]

Hi alfe,

 

But in all honesty I, just like other CA-members, am not sure what you are getting at. I really think it is better if you explain to us in your own words what you try to accomplish so we have the opportunity to help. So far I am not getting the relevance between differences between bit-identical files and, for example, differences (and their cause) between a left- and right channel.

 

 

Kind regards,

Peter

 

I was asked by Superdad about the difference between left and right channel and I answered to it.

As I said before I'm not trying to find any difference between the two files they are identical, what I don't understand and I'm surely wrong, for the last line of noise when I checked the values they are @0,and that's why I asked elsdude to do the test.

[thesurfingalien]

Will do that,thanks

 

Hi alfe,

 

I am not sure if this is what you mean with the zero-bit data, so I am guessing here.

 

If you are referring to a 100% digital silence, AFAIK this is not possible. Since there is an even number of values (2^16 = 65536, for example), there simply is no middle-value. If you generate a "digital silence" in Audacity, you will see signed integer values between +2 and - 3 (if I recall correctly); Dennis (esldude) and I have been discussing this some time ago.

 

If signed integer value 0 indeed *is* the null-line (complete digital silence), then it seems logical that the signed integer range -1 to -32768 would only be used until value -32767 to have an equal number of "positions" above and below to decode the wave-form in. I have tested this, and found it to be incorrect because in several recordings I found where all values (0 to +32767 and -1 to -32768) are used.

 

However, this may not be what you are talking about at all, and it only represents my experiences.

 

Kind regards,

Peter

[esldude]

Alfe,

 

you should have a PM to see what I found. Just looks like normal kind of digital fade out you get when recording digitally form analog tape to me. I sent you the sample values from just before where you specify to the end of the track. They do both go to zero eventually. They fluctuate between zero and the LSB being a 1 for a good many samples. In any case, take a look at the values and maybe we can communicate more clearly about them.

[Jabs1542]

Hi alfe,

 

I am not sure if this is what you mean with the zero-bit data, so I am guessing here.

 

If you are referring to a 100% digital silence, AFAIK this is not possible. Since there is an even number of values (2^16 = 65536, for example), there simply is no middle-value. If you generate a "digital silence" in Audacity, you will see signed integer values between +2 and - 3 (if I recall correctly); Dennis (esldude) and I have been discussing this some time ago.

 

If signed integer value 0 indeed *is* the null-line (complete digital silence), then it seems logical that the signed integer range -1 to -32768 would only be used until value -32767 to have an equal number of "positions" above and below to decode the wave-form in. I have tested this, and found it to be incorrect because in several recordings I found where all values (0 to +32767 and -1 to -32768) are used.

 

However, this may not be what you are talking about at all, and it only represents my experiences.

 

Kind regards,

Peter

 

Peter,

 

Is it possible that this is a "glitch" with Audacity? I could see interpreting the null threshold a minute detail compared to the vast features provided in the software. And the +2 to -3, do you think that's just random low level "noise".

[thesurfingalien]

Peter,

 

Is it possible that this is a "glitch" with Audacity? I could see interpreting the null threshold a minute detail compared to the vast features provided in the software. And the +2 to -3, do you think that's just random low level "noise".

 

Hi Jabs1542,

 

I honestly have no idea... Maybe it is the result of dithering or something; to me it looks completely random. I have tested this behavior with both Linux and Windows versions, with the same results.

 

Unfortunately I have never found an arbitrary document that describes how an "official" digital silence should be defined. If anyone knows about such information, please let me know!

 

Kind regards,

Peter

[esldude]

Hi Jabs1542,

 

I honestly have no idea... Maybe it is the result of dithering or something; to me it looks completely random. I have tested this behavior with both Linux and Windows versions, with the same results.

 

Unfortunately I have never found an arbitrary document that describes how an "official" digital silence should be defined. If anyone knows about such information, please let me know!

 

Kind regards,

Peter

 

Well, redbook specifies 16 bit two's complement. So having zero, plus 32,767 and minus 32.768 is how that would have to work otherwise it isn't two's complement. Don't recall seeing any with plus 32,768, but also wasn't looking for one. When you import something into Audacity if you have it set to use 32 bit float maybe that causes some inconsistency though I am not sure that would work that way either.

[alfe]

Hi Jabs1542,

 

I honestly have no idea... Maybe it is the result of dithering or something; to me it looks completely random. I have tested this behavior with both Linux and Windows versions, with the same results.

 

Unfortunately I have never found an arbitrary document that describes how an "official" digital silence should be defined. If anyone knows about such information, please let me know!

 

Kind regards,

Peter

 

Digital silence - LavryEngineering

 

Audacity Forum • View topic - HowTo: Remove dither noise from silence

[alfe]

@Elsdude

Check your PM

[alfe]

Checked with Elsdude,find out my mistake.

Sorry for bothering you guys with that discussion.

[Audiophile Neuroscience]

So then, 22 pages on and finally we all agree, Alex was right ? Right !

 

Meanwhile, It is chilly since arriving in London but at least I got to enjoy the fabulous sounds of the West End Show, Memphis.Well, I think it was good sound but not sure whether the 'sound guy' measured it so....maybe it wasn't :-). Alex, I bet you would have loved it, if I could put it in an uncompressed zip and send it to you to cheer you up in hospital, I would. When I return I look forward to getting together again to enjoy some music.Get well soon.If I can help in any way just let me know.

 

Cheers

David

[kumakuma]

So then, 22 pages on and finally we all agree, Alex was right ? Right !

 

Am I missing something here? As far as I can tell, we are no closer to figuring out why Alex hears differences in bit identical files.

[RealAudio]

Am I missing something here? As far as I can tell, we are no closer to figuring out why Alex hears differences in bit identical files.

 

 

The answer is simple. Just apply Occam's razor. He is imagining the differences.

[prot]

Am I missing something here? As far as I can tell, we are no closer to figuring out why Alex hears differences in bit identical files.

 

You didnt miss anythin. 22 pages later *everything* is exactly the same. However, looks like the fat lady is still singin

[Audio_ELF]

You didnt miss anythin. 22 pages later *everything* is exactly the same. However, looks like the fat lady is still singin

I'm not sure if Alex listens to opera much :-)

[YashN]

The probable mechanism stays jitter, induced through the ground plane noise.

 

Because the charges in the magnetic storage media are very low signals, there is more or less work needed by the disk electronics to render the bit, causing more ground plane noise, which goes on to affect the threshold at the DAC.

 

When you factor in DRAM, which has leaky charges, and therefore associated refresh mechanisms, things become even more complex.

 

Now, if we assume the power supply quality affects the charges in the HDD, and hence the SQ at the DAC, we should get a similar effect when we copy a file and play it back as compared to the original file.

 

In other words, if you want to test a similar phenomenon to the linear PS rip one, you can test playback of an old file, and a fresh duplicate of that file and listen for differences.

 

It won't be the exact difference if it's there but it is a similar phenomenon (because the electrical activity at the time of the rip is different from the electrical activity at the time of duplication).

 

Short version:

1. Find an old file

2. Duplicate the file

3. Drag and drop each in Bug Head Green

4. Play the original file and listen carefully

5. Play the duplicate file and listen carefully

 

Do blind A/B if it suits your fancy (probably not needed depending on system, abilities).

 

The files should be bit identical.

 

Theoretically the files should have exactly the same SQ when played back.

 

In practice...

 

Am I missing something here? As far as I can tell, we are no closer to figuring out why Alex hears differences in bit identical files.

[jabbr]

The probable mechanism stays jitter, induced through the ground plane noise.

 

Because the charges in the magnetic storage media are very low signals, there is more or less work needed by the disk electronics to render the bit, causing more ground plane noise, which goes on to affect the threshold at the DAC.

 

When you factor in DRAM, which has leaky charges, and therefore associated refresh mechanisms, things become even more complex.

 

Now, if we assume the power supply quality affects the charges in the HDD, and hence the SQ at the DAC, we should get a similar effect when we copy a file and play it back as compared to the original file.

 

In other words, if you want to test a similar phenomenon to the linear PS rip one, you can test playback of an old file, and a fresh duplicate of that file and listen for differences.

 

This is all fine and good theory ... consider however the HQPlayer -> NAA model where the music bits are sent via the galvanically isolated Ethernet first from the NAS to the desktop and then again to a FIFO RAM device (i.e. NAA) and then to the DAC.

 

In this case there is no real mechanism for the magnetic storage ground plane noise to make its way to the DAC.

[YashN]

Yep, most probably, and this is why the server + ethernet + NAA + DAC approach is a very clever one: the DAC isn't exposed as much to the server computer's inherent noise via USB.

 

On the other hand, a computer straight to a USB DAC... many things happen in that system, many very very odd things (which only make some sense if you have read John Swenson's posts around here and AA extensively)...

 

The proxy test above is easy to do (computer to commonly available DAC).

 

If I had a linear PSU, I'd have given it a try too.

 

This is all fine and good theory ... consider however the HQPlayer -> NAA model where the music bits are sent via the galvanically isolated Ethernet first from the NAS to the desktop and then again to a FIFO RAM device (i.e. NAA) and then to the DAC.

 

In this case there is no real mechanism for the magnetic storage ground plane noise to make its way to the DAC.[/color]