100% Natural Sounding Digital Reproduction: Is It Possible???

Transients are waveforms too, so if a transient is within the sampling range of an analog-to-digital convertor, it will get sampled. It doesn't matter how "quick" it is, because a 20kHz sine wave is fast too and IT gets sampled with no problems. Now if that transient is not in the form of a sine wave--which is the only type of wave a digital-to-analog convertor on the playback end knows how to generate--then there will be inaccuracies. But I've never personally heard of a natural sound source NOT emitting sine waves.

I think some people can't properly visualize how sampling actually works because they keep bringing up this "What about the stuff between the sample points?" concept. If that stuff, i.e. sinewaves, are not being picked up by the convertor it's because they are at a higher frequency than the AD convertor was designed to operate at.

BTW: And as far as science not knowing everything, that is true. But it does know a lot of things--I'm pretty sure that Pioneer space probe wouldn't be outside our solar system right now if this wasn't true. So I am not going to adopt this currently fashionable way of thinking that says science is really just a relative thing and everyone's opinion is correct. That's a frightening road that I believe leads to mental chaos and I absolutely will NOT go down that road. I will definitely continue to keep my mind open to new ideas, but they have to have some amount of solid facts to back them up.

 

This is completely, absolutely false. If you hear a difference it is because of distortion in the amp/speakers, not because you can hear the >20Khz components of the square wave.

Let's put this "square wave" nonsense to bed. This is 2004, not 1984.

 

Bob Katz tried an experiment a few years ago of taking high-quality 96K material and filtering out >20Khz (no other changes). Guess what the result was?

 

Sorry but the assumption that we know "everything about anything" is a rather foolish and sometimes very dangerous approach.

How can we measure something if we are not aware of its existence? or...

How can we be sure it doesn't exist just because we may not be aware of it??? Or If science knows all why are people still dying of disease like cancer?

I use my ears to judge the sound quality of music....I suggest others consider taking this approach. I've yet to come across a scientific paper that sounds as good to me as Abbey Road.

Argue theories all you want but, IMHO, with music the proof is in the listening.

 

I am sure that David can defend himself...

but he never said that "we know everything about anything".

 

David, Let me see if I can address some of your concerns. The section of the Sony paper that I suggest people look at is not propaganda, it is pure math. Everyone who understands how digitization and reconstruction of the analog signal knows the results shown in the paper on page 7 are correct. Here's why: Any periodic signal (square, saw tooth, what ever as long as it is periodic) can be constructed from sine wave signals of the same frequency and multiples of that frequency (harmonics). So lets take the 10 KHz square wave (I get to square waves and music in a second) and add the 2nd harmonic to the first harmonic. If you sketch this out you will see that it looks more like a saw tooth wave when you add those two. The third harmonic would help more, it would start to fill out the corners to make the square of the square wave, but the problem is the PCM system sampling at 44.1 KHz has to filter out everything over 22.05 KHz and the third harmonic is 30 KHz. Because of the low sampling rate of the PCM system it does get many building blocks (harmonics to try to make the square wave. On the other hand the DSD system sample 64 time faster than the PCM system. So instead of having just the second harmonic to work with it has all of them out to the 128th harmonic! Thus it does square waves.

Now you are right that music does not have a lot of square waves. Check this link and see what happens with impulse signals comparing PCM and DSD:

http://www.vacuumstate.com/VARIOUS/SACD_2003.pdf

I sure you realize that drum beats and cymbal clashes are basically impulse signals. It is the same situation as the square wave. I hope this helps.

 

Luke, Do have a reference for your claim that this is false, I have played with signal generators going all the way out to 20 KHz using special transducers that do not distort as a typical load speaker does and you do hear the difference.

 

I agree with you. Venue information does appear to be more easily delineated by analogue systems, and digital systems seem to struggle a bit to do that as easily. But the good ones do it, especially when we move beyond Redbook. Analogue struggles a little in other areas that digital just sails through.

 

Don't forget the issue of what type of sound each person likes. I.e. I'm not all that sure if we're all talking about the same thing here. Think about the large variety of speakers all of us own: some people swear by horn-equipped Klipsch designs praising their ability to reproduce every minute musical detail causing a "you are there" sensation........but these same speakers can be deplored by those that own Quad electrostatics who complain that all that detail is much too strident and in your face, but their speakers fill the room with music that has an airy, liquid, and velvety quality that sounds incredibly realistic.......but in response to this, the Klipsch guy says Quads are dull, undynamic, and about as exciting as an unsugared bowl of plain oatmeal.

They are both proclaiming their speakers reproduce music naturally but both people cannot be correct. So until personal opinion is eliminated from the decision process, I think it will be difficult to arrive at an answer to the question of this thread.

So with that in mind, personally I think digital reproduces sound more realistically though analog has the potential to cause it to sound better, i.e. larger than life. That's not necessarily a bad thing because such an ability can cover up problems of other equipment in the preceding signal chain, improper mic placement, bad EQ choices.........whereas digital allows all this to be revealed in all its glory at the playback end >>> that's why I think so many people think digital sounds worse than analog & not because the digital recording process itself inherently sounds bad.

Just one example using something most people here have experience with: cymbals. Lots of people love it when they hear a system reproduce cymbals with a velvety texture. Um, if you've ever heard an actual cymbal played directly in front of you, I wouldn't exactly describe the sound as "velvety". It's more of a sharp, penetrating, highly metallic sound. And when they are given a hard whack, they sound they make can be almost painful to hear. Obviously causing physical pain for the listener is not desirable so something needs to be done to prevent that (EQ manipulation, etc). So we have now entered the messy realm of subjective decision making--in this case, by the mixing engineer, the producer, or the musician himself--as far as what sounds "right". But after this is done, I don't think the cymbal's sound can be considered to be 100% natural anymore.

So to make a decision about what truly sounds natural, I believe the dreaded listening test that uses three letters in its acronym needs to be employed: D.....B.....................

 

I was referring to capturing the exact point in time where a change in amplitude occurs, basically what ALP was talking about with regard to impulses. But like I've said all along, I don't know where all of this falls in the realm of audibility. In other words, does it really matter if the instant the stick strikes the cymbal occurs between sample 1 and sample 2? I don't care all that much because there are so many factors that affect music with the most important, for me, being the mastering. And to answer your other question, nope, my music usually consists of sine waves...except on those days when I feel like a robot. Which is another issue altogether.

 

Perhaps the wisest (and most succinct) thing said on this forum in many a day.

 

Right on. Having played the drums for years I can firmly state that real cymbals sound best with ear plugs.

 

I am sorry David but my experience in the studio has convinced me that much of audio reality is not measurable and I think that's a good thing. I like a little "art" with my "science". Again, I am a big fan of science but I think part of wisdom is admitting when there is no answer. I gave you the MLSSA plots as one example of times in audio where the science came up short in describing phenomena that have a real sonic difference.

Another example is jitter, the time-based distortion in digital audio. Scientists in the audio community where saying that the same DAC should sound the same regardless of transport, after all there was perfect error correction. This was around 1991-1993 time frame. We discovovered, mostly thanks to "audiophiles", that CD players did sound different due to the impact of sound-based distortion. Then the scientists said "aha!" you are right but the human ear can only hear down to thousands of a second. Then audiophiles said "wait, I can hear differences smaller than that". Then Julian Dunn and others discovered sonic differentiation down to nanoseconds. Now there is some belief, I can actually show this on my recording equipment by plugging in a Genex Lucid master clock at 40ps, that picosecond range matters.

My point is this: audio advanacements are real and a very good real example of many times where science cannot perfectly describe what is going on in the audio world.

 

Really? That was my interpretation of his statement that "reality can be measured". If parts of reality cannot be measured, which is what I believe, then we are more on the same page.

Part of science is recognizing its limits.

 

Unfortunately it is even more complex than this. Let's assume (falsely IMHO) that science can measure all (assumption) SIX factors of audio. Even then the interplay of those six factors simultaneously may not be able to be measured. In other words, what I believe based on discussions with physicists at MIT and record engineers that capturing all the complex interactions of audio in real time is almost impossible with today's measuring technology.

As you can infer from my replies, I also think we can't measure all SIX factors to pull a number from the air or that we even know there are SIX factors instead of SIXTY.

Audio is a very highly complex system. I will give you an example...one of my software company consultants co-invented the "digital Strad" cello made famous at MIT's Media Lab. There was a need to replicate this cello by taking in changes in bow speed and other dimensions of playing a cello (he is also an accomplished cellist in addition to being a top mathematician). At first he thought the hard part was to build an instrument that could capture all the different arm movements, fingerings, bowing, etc. involved in playing an instrument. What turned out to be different, however, was the billions plus data points per second that the sensors picked up. They literally invented in a way a new class of mathematics to handle the large amount of data produced. Even then, the digital cello was far from giving the tonality of a Strad cello.

Now imagine an orchestra of similar and different instruments...isn't it a bit arrogant in a way to suggest we can measure every aspect of that entire performance?

 

no recording can ever be 100% natural sounding.

the issue is which is most musical, has the most lifelike color, depth, spacial imaging, etc.

but it will always be a recording.

SACD 5 channel with a good recording and good mix and good mastering sounds pretty natural to these ears. But Vinyl on an audiophile setup is the most pleasing/musical. However, it's an artificial experience, in 2-channel stereo. It's not "natural", just less stepped and more flowing and warm.

I think the spacial characteristics of the waves in a 5-channel SACD mix are closer to nature.

if you are asking "can any digital recording not sound like synthesis?" then that is a good question. It beats me! Even SACD sounds like synthesis in my experience, but then maybe if you spent a LOT of money with all balanced Macintosh amps and good vintage speakers and a $3500 SACD player, then you would find that you would not hear synthesis at all.

However, most of us don't have that kind of money. So it's a trade-off - there isn't really 5-channel vinyl.

Maybe there could be a tape format but who would buy it??? (Besides me...)

 

I think Steve has it right. DSD is the closest format to natural realistic sound we can get right now. Something better (maybe a higher sampling rate based on what I am hearing) will come along as bandwidth gets cheaper and more manageable.

 

I'm not so sure. In my mind, hearing is an analog process as neurons fire differently as the tympani of the ear drum hear sounds. And aren't sound waves analog in nature? There's no real digital to analog or analog to digital process is there?

 

http://world.std.com/~griesngr/intermod.ppt

"THE essential paper on this subject is: Karou and Shogo “Detection of Threshold for tones above 22kHz.” – Convention paper 5401 presented at the 110th Convention, May 12-15 2001, Amsterdam.
The authors presented 13 subjects with a test signal consisting of a 2kHz tone combined with odd order harmonics, both sonic and ultrasonic.
The ultrasonic harmonics were switched on and off at a 2Hz rate.
ALL subjects could discriminate the ultrasonics when the combined signal was presented through a single loudspeaker.
NONE of the subjects could discriminate the ultrasonics when each ultrasonic harmonic was reproduced from a separate speaker."

 

Luke, You have confused me. This confirms my claim. That humans can hear the ultrasonic contribution to a sonic fundamental even though they cannot hear the ultrasonic when it is the fundamental. Luke, At what frequency to you think you can hear the difference between a sine wave and a square wave: 100 Hz, 1 KHz, 2 KHz, 5 KHz ....

 

Yes, you're confused. The ultrasonics are inaudible by themselves or in combination with audible tones - when the ultrasonic component is reproduced using a seperate amp/speaker. If you use a single amp/speaker, then you may hear a difference, but the reason is not ultrasonic hearing.

 

Luke, You confused me but the article does not, it is saying exactly the same thing that I am saying. Humans cannot hear ultrasonic frequencies when they are the fundamental frequency. It does not matter if the ultrasonic is being played by itself or a sonic fundamental is be played through another amp/speaker at the same time. Humans do not hear ultrasonics when they are the fundamental, HOWEVER (as the article says), humans can hear the effect of the ultrasonics on a sonic fundamental when they two signals are combined into one signal and played through a single amp/speaker. It has nothing to do with distortion in the amp/speaker, it has do with the fact that the ultrasonics change the shape of the sonic signal (in particular the leading edge of the signal) and we can hear that.
The author and I disagree about why people hear the difference, but not the fact that people hear a difference.

 

I'm not approaching this from any kind of engineering point of view, simply as a listener of hi-fi. Personally, I think my EMM Labs digital gear is as close to an "analog" sound as I've ever heard from a digital component, without losing any of the benefits of digital reproduction. Whether it mimics the analog waveform perfectly is moot to me. I'm having too much fun listening.