Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Zep Dude]

Hi Dan, I recently did some listening of various AD's: Lavry Gold, Mytek, Benchmark, Apogee AD16x, Rosetta 800.  I passed 2 track analog masters containing full mixes as well as individual instruments through each unit.

The differences were most obvious between the units with full mixes.  With individual instruments things become more subtle.

I have asked some techs who have agreed with the idea that it is more challenging for the converter to convert complex sonic material containing dense harmonic content verses single instruments.

Would you agree with this?

[Jon Hodgson]

Zep Dude wrote on Thu, 20 October 2005 10:44

Hi Dan, I recently did some listening of various AD's: Lavry Gold, Mytek, Benchmark, Apogee AD16x, Rosetta 800.  I passed 2 track analog masters containing full mixes as well as individual instruments through each unit. The differences were most obvious between the units with full mixes.  With individual instruments things become more subtle. I have asked some techs who have agreed with the idea that it is more challenging for the converter to convert complex sonic material containing dense harmonic content verses single instruments. Would you agree with this?

No.

Though if your analogue stages have some non-linearities in them then you're going to get a lot more different sum and difference frequencies with a full mix, which might be less pleasant. The converter itself I would not expect to have a problem.

But Dan's the one who actually builds converters, so hopefully he'll chime in and tell me if I missed something.

[danlavry]

Jon Hodgson wrote on Thu, 20 October 2005 13:03

Zep Dude wrote on Thu, 20 October 2005 10:44 Hi Dan, I recently did some listening of various AD's: Lavry Gold, Mytek, Benchmark, Apogee AD16x, Rosetta 800.  I passed 2 track analog masters containing full mixes as well as individual instruments through each unit. The differences were most obvious between the units with full mixes.  With individual instruments things become more subtle. I have asked some techs who have agreed with the idea that it is more challenging for the converter to convert complex sonic material containing dense harmonic content verses single instruments. Would you agree with this? No. Though if your analogue stages have some non-linearities in them then you're going to get a lot more different sum and difference frequencies with a full mix, which might be less pleasant. The converter itself I would not expect to have a problem. But Dan's the one who actually builds converters, so hopefully he'll chime in and tell me if I missed something.

That is a tough question. I can see some mechanisms, however subtle, so I would not brush the issue aside. In other words, I need time to think about it.

There are mechanisms to explain differences between solo instrument against the same instrument in the presence of other instruments. One such mechanism is about integral non linearity. Compare a solo flute occupying say 10% of a converter range, against the same flute with say a string section. With the strings, you can view the flute as if it rides on the "rest of the sounds". So in fact, some lower frequencies can move the "baseline" of the flute from one "section" of the converter to another. A non linear transfer curve (integral non linearity) will modulate both the amplitude and the distortions of the flute, as it causes it to "exercise" different "regions". Of course one can view the flute as modulating the strings, and in fact, an integral non linearity makes everything interact in an undesirable fashion.

Other mechanisms are even more complex, and are signal dependent component variations, signal dependent power supply variations and more. True, such variations would happen with solo instrument, but in a "static" manner. A constant amplitude and a constant distortion on a solo instrument is less offensive then a time varying change, because the ear is much more sensitive to changes. For example, if I change the amplitude by a fixed .1dB, you may not know it. But change it back and forth at 1KHz, and you will know it right away...

How much non linearity is audible? How much in signal dependent component variations? I do not know. I just try to minimize those effects the best I can. I did perform some studies regarding capacitor memory effects (dielectric absorption),  signal impact on resistors and semiconductors. But I never "quantified the results against the ear". I do not know anyone that has done so. It would require a ton of blind ABX listening tests.

Regards
Dan Lavry
www.lavryengineering.com

[Jon Hodgson]

danlavry wrote on Fri, 21 October 2005 00:09

Jon Hodgson wrote on Thu, 20 October 2005 13:03 Zep Dude wrote on Thu, 20 October 2005 10:44 Hi Dan, I recently did some listening of various AD's: Lavry Gold, Mytek, Benchmark, Apogee AD16x, Rosetta 800.  I passed 2 track analog masters containing full mixes as well as individual instruments through each unit. The differences were most obvious between the units with full mixes.  With individual instruments things become more subtle. I have asked some techs who have agreed with the idea that it is more challenging for the converter to convert complex sonic material containing dense harmonic content verses single instruments. Would you agree with this? No. Though if your analogue stages have some non-linearities in them then you're going to get a lot more different sum and difference frequencies with a full mix, which might be less pleasant. The converter itself I would not expect to have a problem. But Dan's the one who actually builds converters, so hopefully he'll chime in and tell me if I missed something. That is a tough question. I can see some mechanisms, however subtle, so I would not brush the issue aside. In other words, I need time to think about it. There are mechanisms to explain differences between solo instrument against the same instrument in the presence of other instruments. One such mechanism is about integral non linearity. Compare a solo flute occupying say 10% of a converter range, against the same flute with say a string section. With the strings, you can view the flute as if it rides on the "rest of the sounds". So in fact, some lower frequencies can move the "baseline" of the flute from one "section" of the converter to another. A non linear transfer curve (integral non linearity) will modulate both the amplitude and the distortions of the flute, as it causes it to "exercise" different "regions". Of course one can view the flute as modulating the strings, and in fact, an integral non linearity makes everything interact in an undesirable fashion. Other mechanisms are even more complex, and are signal dependent component variations, signal dependent power supply variations and more. True, such variations would happen with solo instrument, but in a "static" manner. A constant amplitude and a constant distortion on a solo instrument is less offensive then a time varying change, because the ear is much more sensitive to changes. For example, if I change the amplitude by a fixed .1dB, you may not know it. But change it back and forth at 1KHz, and you will know it right away... How much non linearity is audible? How much in signal dependent component variations? I do not know. I just try to minimize those effects the best I can. I did perform some studies regarding capacitor memory effects (dielectric absorption),  signal impact on resistors and semiconductors. But I never "quantified the results against the ear". I do not know anyone that has done so. It would require a ton of blind ABX listening tests. Regards Dan Lavry www.lavryengineering.com

Hi Dan.

Thanks for the explanation.

If I understand your points correctly I was on the right track, though I was probably rather too abrupt in my statement. I'm afraid I'm a little worn out by people trying to find fault in the basic sampling theorem.

So Zep Dude, what I should have said is that I believe there is nothing INHERENT in sampling which would mean that multi-instrument mixes would be more challenging to capture than a single instrument one, however there are various ways in which signal dependent non-linearities can occur in a real world converter system (which is largely analogue circuitry), and more complex signals could generate more complex and less pleasant distortions.

I had only been thinking about the harmonic content of that distortion, but as Dan points out, in a mix the combinations of sounds is probably more dynamic than with a solo instrument, which would lead to distortion content which changes, and our ears and brains are better at picking out changes than constants.

So sorry Zep Dude if I was a little abrupt in my rebuttal.

And thanks again Dan.

regards

Jon

[Zep Dude]

I don't want to put this topic to bed quite yet because IF it is true, and Dan is hedging that it may, there are many implications in terms of every day usage and selection of converters and even arguements to be made about analog summing:

These are some potential "rules" that could be drawn from this if it were true:

1.  Assuming AD converter "A" sounds noticably worse than converter "B" when converting an entire mix, converter A may still approach converter B in overall sound quality if tracking that song one instrument at a time.  Thus a final mix made from tracks converted with the inferior converter A  might sound very similar in quality to an identical final mix using the sum of tracks converted with superior converter B.

2.  The reason that Analog summing can sound better than mixes done "in the box" might have nothing to do with digital summing in a DAW but really the fact that in creating "stems" and sending them out of seperate AD outputs, one is really making it easier on the AD converters to perform their tasks well.  This would explain why some people hear a difference when analog summing while others do not.  People with top quality converters would be less likely to hear improvement, while people with lesser quality converters would gain more because those converters are unable to pass full mixes well but can do fine with smaller stems.

Dan, are you still out there?

[Jon Hodgson]

Zep Dude wrote on Fri, 21 October 2005 06:03

I don't want to put this topic to bed quite yet because IF it is true, and Dan is hedging that it may, there are many implications in terms of every day usage and selection of converters and even arguements to be made about analog summing: These are some potential "rules" that could be drawn from this if it were true: 1.  Assuming AD converter "A" sounds noticably worse than converter "B" when converting an entire mix, converter A may still approach converter B in overall sound quality if tracking that song one instrument at a time.  Thus a final mix made from tracks converted with the inferior converter A  might sound very similar in quality to an identical final mix using the sum of tracks converted with superior converter B. 2.  The reason that Analog summing can sound better than mixes done "in the box" might have nothing to do with digital summing in a DAW but really the fact that in creating "stems" and sending them out of seperate AD outputs, one is really making it easier on the AD converters to perform their tasks well.  This would explain why some people hear a difference when analog summing while others do not.  People with top quality converters would be less likely to hear improvement, while people with lesser quality converters would gain more because those converters are unable to pass full mixes well but can do fine with smaller stems. Dan, are you still out there?

I find your logic good on both points.

However since this non-linearity is essentially a problem in the analogue domain, and most of the kinds of artifacts we see should also occur in pure analogue equipment to varying degrees, then we should notice similar variations in solo/mix performance between different desks, preamps, mix busses etc.

So the question is, in the top end (i.e. high quality units) do we hear those differences?

[Sam Lord]

Just my 2 cents: the market for symphonic CDs compared with chamber music CDs has dropped at lot, at least in this household.  At 16 bits output, no matter the fidelity of the recording chain, very complex acoustic music doesn't cut it period, IMO.  At 24 bits, the problems are apparently gone.  For solo artists or very processed music, I find the difference between well-done 16-bit and hi-res very slim.  With big choirs and bands, the difference slaps you.  Thanks for listening...  

[maxdimario]

the inner detail in music gets progressively more difficult to reproduce as the number of sounds in the arrangement increase and the waveform becomes more complex and closer to 'chaotic'.

I totally agree about the difference between symphonic vs. chamber (and solo instruments -- which sound pretty good compared to full orchestras) in cd releases above.

whether this is a question of the analog driver for the DAC, the bit depth/sample rate, or the quality of the converter overall I believe the answer is yes to all.

for instance the 'bat out of hell' cd which is saturated with instruments has a difficult time keeping the separation between instruments on the cd, even though dynamic range is limited. The vinyl from the 70's does not have this problem.

'should not' doesn't explain it.

[Jon Hodgson]

maxdimario wrote on Fri, 04 November 2005 08:26

The vinyl from the 70's does not have this problem.

Well then bit depth and sample rate are not the problem, unless the problem is TOO MUCH bandwidth or a noise floor that is TOO LOW.

I would ask a couple of questions...

1) Was the album remixed for CD, or did it come from the same 2 track masters?

2) What effects do the cutting lathe, its drive circuitry, and your record player have on the signal?

[Yannick Willox]

I do not agree on the symphonic vs chamber argument.
One of the arguments I originally got into mobile recording was that a lot of solo/chamber music cds just sound awful (compared to the real thing). Most (or at least more) orchestral cds seem to sound more or less as an orchestra.

You can compare it to editing a cd : chamber music 100 to 500 edits, an orchestral recording gets away with a lot more mistakes/problems, hence less edits (not to forget the cost issue to do enough takes).

A solo instrument is so nude that you can hear every defect imposed by the recording/playback chain.
An orchestra - by it's complexity - tends to mask several problems.

I would expect an AD/DA converter to be more exposed by a single, harmonically rich instrument, such as harpsichord or a 17th century violin : the harmonics tend to be MUCH louder than a modern(ised) instrument, so intermodulation distortion should be much clearer than on complex recordings, where you would get a lot of masking.

Also the cleannes of the attacks can be evaluated much more clearly on small ensembles/solo recordings.

Just try to get an orchestra to -really- play together, so you can evaluate transient response ...

[maxdimario]

I think it's more of a question of masking effects.

the orchestra can be in or out of sync, but the individual instruments can lose their 'identity' and some instruments have the effect of getting lost...difficult to pinpoint and single out.

this can be devastating to a mix where there are so many things fighting each other that even the slightest problem with masking can upset the balance.

digital ITB arrangements (for example)sound best when there's fewer, bigger sounds..as opposed to heavy layering...but that is only my personal observation and a few friends'...can't be proven unless someone does a survey or something.

strangely enough, vinyl had the opposite effect: solo acoustic GTR. for instance sounds slightly thin (even though I personally enjoy the high freq. range more)on vinyl... and so low speed disc (33.3 rpm LP) tends to sound more convincing when you add more instrumentation than you would if you were monitoring straight off the mixer, to make the records fuller. My observation, again..