Depth, the final frontier.

[bdiament]

Hi jkeny,

...I guess you could do the test with the MH then but why bother - I'm sure you will find bit-perfectness between the two playbacks?...

Sorry I wasn't clear. You stated "Nearly all audio DAC chips, AFAIK, take I2S signalling as their input & this could be captured" so I was curious as to whether you'd actually experienced this. I know it can be done with MH gear (I mentioned having done it in an earlier post) and was wondering, based on your statement, what you had used to do the same.

...My point is that you state you can hear differences between PCM & lossless file playback. These produce equivalent bit-perfect datastreams, right? So what is causing the difference in sound between these two playbacks?

I don't know if they produce equivalent bitstreams. If they do, I don't know what is causing the sonic discrepancies I hear.
The only difference is the additional layer of encoding/decoding so I suspect the answer(s) lie in that direction.

I'm much less concerned about what is causing the differences than the fact that I hear the differences. As far as I'm concerned, hard drive space is cheap and I see no point in reducing the size of a file. I don't reduce the size of files when mixing for clients and I don't reduce the size of files when mastering - both of which involve considerably more computing horsepower than simply listening to the finished results. Personally, I much prefer to leave the files in raw PCM format, so that's what I do. I'll leave the vicissitudes of file size reduction to the folks who are concerned with such.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

[jkeny]

Hi jkeny,
.....................
I don't know if they produce equivalent bitstreams.

Really? I'm sure you know the concept behind lossless encoding!

If they do, I don't know what is causing the sonic discrepancies I hear.
The only difference is the additional layer of encoding/decoding so I suspect the answer(s) lie in that direction.

............
Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

OK, but that doesn't answer the question I'm posing - what in the digital datastream which hits the DAC chip is different between the two playbacks, PCM & Lossless - it's not the bits so what then ..........

 

[bdiament]

Hi jkeny,

Really? I'm sure you know the concept behind lossless encoding!

I know what the *theory* states but as Yogi Berra said "In theory, there is no difference between theory and practice. In practice, there is." ;-}

OK, but that doesn't answer the question I'm posing - what in the digital datastream which hits the DAC chip is different between the two playbacks, PCM & Lossless - it's not the bits so what then ..........

My answer was in an earlier post: I don't know. Do you have an answer?
I'm open to any ideas, though as I said, I'm less interested in the "why" in this case than in the "what".

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

[opus111]

If the data *entering the DAC* is the same, I cannot imagine any mechanism by which the output of the DAC would differ. (Of course, that doesn't mean such a mechanism might not exist - just that at this point, I can't imagine what such might be.)

In my experience, its most commonly what's called 'common ground impedance coupling'. Put another way, dirty grounds. In the case of USB, a common ground connection between the PC and the DAC can easily result in a ground loop (via coupling capacitance to mains via transformers in PSUs) at the very high frequencies involved wherever PCs are concerned.

 

[jkeny]

In my experience, its most commonly what's called 'common ground impedance coupling'. Put another way, dirty grounds. In the case of USB, a common ground connection between the PC and the DAC can easily result in a ground loop (via coupling capacitance to mains via transformers in PSUs) at the very high frequencies involved wherever PCs are concerned.

Yes & in my experience it could also have a lot to do with the integrity of the USB signal waveform & possibly the timing of the packetised nature of the USB protocol.
The audio industry is only waking up to this now & is at the start of investigation into this. The above guesses may prove to be incorrect, it really doesn't matter, what matters is that in digital audio it isn't just the bits that are important, it's the low-level electrical representation of these bits & the carrier environment that needs to be scrutinised/examined & studied.

 

[bdiament]

Hi opus111,

In my experience, its most commonly what's called 'common ground impedance coupling'. Put another way, dirty grounds. In the case of USB, a common ground connection between the PC and the DAC can easily result in a ground loop (via coupling capacitance to mains via transformers in PSUs) at the very high frequencies involved wherever PCs are concerned.

If the data is following a single path - in my case, from the hard drive, via Firewire, to the DAC - then wouldn't any common ground impedance coupling affect everything taking that path in the same way? (Note there is no USB involved.)

If it does *not* affect anything taking that path in the same way, I would think any affects would then have to be random, in which case, what sounds "better" would vary. And if it *does* affect everything the same way, I don't understand how any audible differences could be attributed to it.


Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

[opus111]

Hi Barry

If the data is following a single path - in my case, from the hard drive, via Firewire, to the DAC - then wouldn't any common ground impedance coupling affect everything taking that path in the same way? (Note there is no USB involved.)

I'm not clear here what you are asking. Are you imagining that this common ground impedance is affecting the data? If so no, the error is not of sufficient magnitude to flip a data bit.

I am guessing (but don't know for sure as I'm unfamiliar with Firewire) that there's a common ground connection, just as with USB. Once there's a loop though then everything sitting in the path of that current loop is disturbed. Digital stuff is immune to the relatively small disturbances involved, but analog stuff not. So then if your DAC is connected to a mains supply, its internal 0V (gnd) will be a path for the noise currents unless the Firewire interface is totally isolated.

 

[jkeny]

Sorry for my use of USB as the example. To generalise, it would appear from my preliminary experiments that the digital signal waveform has some effect on the resultant audio. whether this is noise born interference or waveform distortion (none of which will cause bit error) or some other characteristics I don't know. The important thing to note is that a lot of interference can exist on the digital signal which will not cause any problem once we remain in the digital domain but once we step into the analogue domain, it's a different set of considerations.

 

[bdiament]

Hi opus111,

Hi Barry

I'm not clear here what you are asking. Are you imagining that this common ground impedance is affecting the data? If so no, the error is not of sufficient magnitude to flip a data bit.

I am guessing (but don't know for sure as I'm unfamiliar with Firewire) that there's a common ground connection, just as with USB. Once there's a loop though then everything sitting in the path of that current loop is disturbed. Digital stuff is immune to the relatively small disturbances involved, but analog stuff not. So then if your DAC is connected to a mains supply, its internal 0V (gnd) will be a path for the noise currents unless the Firewire interface is totally isolated.

Perhaps I misread. I thought *you* were suggesting that common ground impedance was resulting in an affect on the *audio*.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

[jkeny]

Hi opus111,

Perhaps I misread. I thought *you* were suggesting that common ground impedance was resulting in an affect on the *audio*.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

I believe he was Putting it simply, I believe, dirty grounds can effect the the D/A conversion & therefore the quality of the audio output!
I think you are assuming that the only way the audio can be effected is that the digital data has bit errors in it - i.e it's not bit-perfect? Correct me if I'm wrong!

 

[bdiament]

Hi jkeny,

I believe he was Putting it simply, I believe, dirty grounds can effect the the D/A conversion & therefore the quality of the audio output!

If so, then my question from post #106 still stands.

I think you are assuming that the only way the audio can be effected is that the digital data has bit errors in it - i.e it's not bit-perfect? Correct me if I'm wrong!

Your assumption about what I might be assuming is indeed incorrect. I am not making any assumptions at all, particularly with regard to digital audio.

What I've said is, if, in the digital domain, I can null two files against each other, 100% to the sample, that *suggests* to me what what I'm feeding my DAC is identical in both instances. I would expect them to be treated identically by the DAC chip, audio-wise, jitter-wise and any other way I can imagine. Ergo, I would expect the output from the DAC chip to the analog stages and ultimately, the output from the box to my power amps, to be identical too.

To be clear, I'm not saying this IS the case and I'm not saying it is a Universal Truth (please add reverb to those words ;-}). I'm saying this is my understanding and expectation, flawed though they may be in the face of any new evidence to say otherwise.

Not at all the same (at least to me) as "bits is bits" because the latter doesn't take into account timing between those bits during decoding and perhaps a host of other as yet unquantified factors. I've certainly heard differences, from very subtle to quite pronounced, in comparing CD pressings from different replication facilities against each other and against the master used to create them. The data, once extracted to computer, have consistently (i.e. 100% of the time over several thousand examples in the course of close to three decades now) proven to be 100%, bit-for-bit identical. Of course, except in the case when one is in the presence of a fax machine, we don't listen to data.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

[opus111]

If so, then my question from post #106 still stands.

That's the question I couldn't understand. Particularly the 'everything taking that path' bit.

What I've said is, if, in the digital domain, I can null two files against each other, 100% to the sample, that *suggests* to me what what I'm feeding my DAC is identical in both instances.

Seems a bit stronger than suggest, assuming reasonably competent engineering of course.

I would expect them to be treated identically by the DAC chip, audio-wise, jitter-wise and any other way I can imagine. Ergo, I would expect the output from the DAC chip to the analog stages and ultimately, the output from the box to my power amps, to be identical too.

Yep, but the output from the DAC chip is coupled into the following analog stages via PCB traces that may well be affected by noise currents in the ground. Some DACs have differential outputs, signals from these will be relatively immune whilst the signal is in differential form - however once its converted back to single ended (gnd referenced) it becomes susceptible again.

 

[bdiament]

Hi opus111,

That's the question I couldn't understand. Particularly the 'everything taking that path' bit.

What I was getting at is that any noise or other artifacts engendered by passing the signal through my DAC (meaning input receiver, through DAC through analog stages through output - the whole enchilada), would (at least I expect they would) have the same affects on any signal I sent through that path. If this is the case, even if said affects were somewhat randomized, I don't understand why certain formats, i.e., raw PCM such as .aif or .wav, would consistently sound indistinguishable (to me) from the master while other formats, i.e. so-called "lossless" formats, when expanded during listening, would consistently sound different from the master.

Perhaps I'm mistaken (always a possibility I leave room for) but I don't see noise currents from ground as the reason for this. Unless certain formats are inherently more susceptible to such but I don't see that either. Again, I could be wrong.

Seems a bit stronger than suggest, assuming reasonably competent engineering of course.

To me, "suggests" feels reasonable, in that I don't want to make any assumptions. ;-}

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

[opus111]

What I was getting at is that any noise or other artifacts engendered by passing the signal through my DAC (meaning input receiver, through DAC through analog stages through output - the whole enchilada), would (at least I expect they would) have the same affects on any signal I sent through that path. If this is the case, even if said affects were somewhat randomized, I don't understand why certain formats, i.e., raw PCM such as .aif or .wav, would consistently sound indistinguishable (to me) from the master while other formats, i.e. so-called "lossless" formats, when expanded during listening, would consistently sound different from the master.

Ah, OK - elucidation appreciated.

Well the hypothesis (and its a very tentative one at this stage) is that uncompressed formats (e.g. .wav) tax the CPU less than compressed (e.g. FLAC) ones. So the noise is different between the two cases. More CPU effort results in more noise and more signal corruption post-DAC. Thus 'lossless' formats would sound less transparent under this proposed mechanism.

 

[bdiament]

Hi opus111,

Ah, OK - elucidation appreciated.

Well the hypothesis (and its a very tentative one at this stage) is that uncompressed formats (e.g. .wav) tax the CPU less than compressed (e.g. FLAC) ones. So the noise is different between the two cases. More CPU effort results in more noise and more signal corruption post-DAC. Thus 'lossless' formats would sound less transparent under this proposed mechanism.

Okay, now I understand your point, thanks. I think it will be interesting if this turns out to indeed be the case.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

[Phelonious Ponk]

Hi opus111,



Okay, now I understand your point, thanks. I think it will be interesting if this turns out to indeed be the case.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

How would one test this hypothesis? Measure the noise levels, after conversion to analog, of lossless vs. uncompressed formats? Doesn't seem like this would be that difficult. And if there are differences, but they're very small, then you would need to test audibility, blind, see if someone can hear the difference when they're not expecting a difference.

It all seems to come back to the same place, over and over again.

Tim

 

[opus111]

How would one test this hypothesis? Measure the noise levels, after conversion to analog, of lossless vs. uncompressed formats?

That does seem rather difficult seeing as we're not quite sure where to look for the noise. Also noise is very difficult to quantify, by nature. I'd prefer a slightly more indirect method - implement a DAC where common-mode ground impedance coupling has been eliminated by design. Then do blind listening tests on compressed vs uncompressed formats and see if anyone notices one's better than the other.

Alternatively, to some degree common-mode noise can be mitigated by use of clamp-on ferrites (lots of them!) around cables. I know USB is differential (dunno about Firewire) so this won't suffer from having plenty of ferrite wrapped around it. Then again, blind listening to note if it becomes progressively more difficult to tell apart the formats as more ferrite (aka filtering) is added.

 

[bdiament]

Hi Tim,

How would one test this hypothesis? Measure the noise levels, after conversion to analog, of lossless vs. uncompressed formats? Doesn't seem like this would be that difficult. And if there are differences, but they're very small, then you would need to test audibility, blind, see if someone can hear the difference when they're not expecting a difference.

It all seems to come back to the same place, over and over again.

Tim

We haven't tested the noise levels but we did conduct blind comparisons here back when we were first considering Soundkeeper's high res formats and how we were going to distribute them. In our tests, everyone consistently picked the "lossless" file every time. It didn't seem like a very subtle difference when put up against the raw PCM high res files.

(We also considered Blu-ray but for other reasons, decided to pass.)

Yet, some folks say they find "lossless" indistinguishable from raw PCM. Others tell me they hear .wav as sounding "better" than .aif - a difference I have not discerned but not for lack of trying. Whether all this is attributable to different systems or different listeners or both, I don't know.

Perhaps Flac sounds exactly like .aif (though it doesn't to me) and .wav is indeed better than .aif (though I can't tell them apart when listening).
I can only go by what my ears and those of folks I trust who shared in the comparisons here, tell me.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

[Phelonious Ponk]

Not sure I completely understand the difference between raw PCM and .wav or .aif, Barry.

Tim

 

[bdiament]

Hi Tim,

Not sure I completely understand the difference between raw PCM and .wav or .aif, Barry.

Tim

Sorry if I wasn't clear. By "raw PCM" I'm referring to .aif or .wav format.
My statement about some folks telling me they find .wav "better" than .aif was just to list another area where I've heard folks reporting on sound differences. I did not mean it to somehow suggest .aif or .wav are something other than the raw PCM they are. (In other words, by "raw", I mean a PCM format that has not been reduced in any way.)

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com