A violin has to sound like a violin

[syganymede]

There is considerable debate about HiFi quality. And one of the most frequent misconceptions is that since the human ear can hear at very best 20-20kHz this is all we need. And this is provided by CDs. But it is not so.

 

The aim of a HiFi system is to capture the sound made by the instruments and then reproduce this. But instruments do NOT stop at 20kHz as many 24bit/96kHz recorded spectrums will show. In fact on most of them I have looked at there is considerable frequency content up to at least 35kHz. (a 96kHz recording can reproduce, at best 96/2 due to the Nyquist theorem, so has a max bandwidth of 48kHz. Just as the 16bit/44.1kHz has a maximum of 22kHz).

 

So we are faced with two problems. First to be able to record and reproduce sounds way outside our hearing if we want to reproduce the realistic sound of many instruments. Second how to get recording studios to deliver this sound.

 

We need better equipment, capable of storing big files and with much larger bandwidths for playback. At least to 40kHz or so to cover the sounds the music instruments generate. We should use 24bit, not 16, as we need the dynamic range to accurately reproduce the "dying" sounds - like the end of a cymbal crash, or the breath across the flute.

 

I believe that we need a technology step-up from CDs to a better general medium for delivery of mainstream music. This can be downloads or DVDs. We need to think that 24bit/96kHz (FLAC or AIFF or WAV, but uncompressed, ie NOT MP3 or AAC) as the main stream standard, not 16/44.1kHz which is now years old. This, as yet, does not seem to be coming from the recording industry. Maybe Apple iTunes could start an audiophile sections for HD downloads and get the studios moving?

 

See a full analysis at http://www.syganymede.eu/Ganymede/Audio_matters_files/Analysis%20of%20digital%20music.pdf

 

[PeterSt]

Supposed I would like to follow your reasoning, it is worse than you think. Why ? because your 20KHz (or 22 for that matter) can not reproduce any form of a sine which could be in the data. 22050Hz is a pure square, no matter the original was a pure sine. 11025 Hz is a two step square, and so on.

The only thing happening at doubling the sample rate, is that the 22050Hz will be come a two-step instead of a one-step square. Of course this is better, but still miles off. However :

 

Personally I don't have problems with this at all, and since I even think non-oversampling and filterless sounds the best (compared to upsampled), the above is theory only.

 

For me it doesn't need any fix. You're far far better off with a good DAC and good playback software. I can tell this by listening, which is a kind of opposite from following theories.

 

My 2c,

Peter

 

[audioengr]

24/96 can reproduce a realistic violin. I've heard it. The caveat is that the digital system must:

 

1) have very low jitter

2) very low noise

3) very low distortion

 

I believe that the recording side needs more work frankly. Some are finally doing a great job. I think what is being accomplished at Bluecoastrecordings demonstrates what is possible at 24/96. Some of these recordings are so live in my system, my dog barks at it. I have to do double-takes on whether someone is in the room. It's truly startling sometimes.

 

It is very difficult to record orchestral pieces in large venues. Even a piano is a major challenge, not to mention a kick-drum.

 

Steve N.

Empirical Audio

 

[sshd]

In order to take high definition content seriously, a system able of reproducing these frequencies is needed. Soundcard is easy. Amplifier may take a little looking around but should not be a problem either. But for speakers you need something very special and very expensive. Most speakers (tweeters) cut off around 20.000.

 

If you got the proper system, a simple test can whether these unhearable frequencies can be perceived and affect the sound. Compare these three blindly:

A: Some high definition recording with a violin

B: Same recording with high frequencies cut off

C: Same recording with falsified high frequencies (play them backwards or take some from a completely different recording).

 

I am unable to make such a test myself because I lack the speakers.

 

[PeterSt]

I'm afraid, like the OP, you are looking at the wrong phenomena. I don't say it is easy to understand, but here's an attempt to make things clear better :

 

First off, one may think that higher frequencies can be perceived in some direct or indirect manner. Well, that might even be so, but it is not what the difference makes ... if we could perceive that at all. The point is, before it is going to happen that we might perceive those frequencies, first a few other things have to be arranged for (besides a properly resolving system of course !);

 

A main difference between playing back 44K1 vs. e.g. 96K is that Nyquist is shifted. This comes down to what I said earlier in this thread, and no matter I reason that the difference isn't all that large for proper (sine) frequencies, it is the huge amount of harmonic distortion coming from those false sines that is shifted out of the audible band (still only to some extend, but it matters for sure).

Thus, the sole fact that you play at a higher sample rate, makes SQ better.

 

But there is much much more going on, and I am sure I am going to loose you now :

 

I am working on these things for too long not to know that any random oversampling DAC is not able to represent violins (a.o.). This is because the net effect of an OS DAC creates just the opposite of what I said earlier : all genuin square(ish) waves are molested into pure sines.

From this point on (using an OS DAC) all is lost, and it really doesn't matter whether you're using 44K1 or 96K.

 

Keep in mind : square waves create harmonics, whereas sines do not. And thus, any instrument or voice containing squareish information hence imply harmonics, come out as sines without harmonics.

A first intrument to suffer from that is the violin, and when your system doesn't sound harsh in the first place (like with poor PC playback) it will turn into a fluteish instrument.

Now you know this, go out and compare OS to NOS, and you will hear this immediately.

 

So, we use NOS DACs from now on, right ?

I'm afraid this is way more complex (for comparison) than just that. This is because the NOS DAC just creates opposite problems; just read back my first post here, and read again about a 20KHz tone being a pure square wave, and ... see that with the NOS (filterless) DAC this now just stays so ...

This, while the OS DAC can (no, will) create nice sines from that again.

 

Now add to this that the NOS/filterless DAC will have squares where sines were intended, and know that you will be ending up with a huge pile of harmonic distortion because of that.

 

In the end the whole point is that it is either this or that, and there's nothing in between. Both is bad. However :

 

Loosing all harmonics in the first place is just asking for poor instrument representation (a flute can stay), while incurring for harmonic distortion always gives the chance of that not being audible. And you know, it just is not. It is not, probably because the amplitudes of it are always under the original levels, or otherwise this is still to find out (I am working on this myself explicitly).

 

Of course I am a bit speculative on those thinking like this about violins will be listening to OS DACs.

Also, playing a higher samplerate does matter a little, because there's just one step less of turning squares into sines (you start off higher) hence there's just a little less flattening of harmonics.

 

Lastly I can tell you that quite a few people dropped by here, all being convinced of OS being "it". Without exception all were converted within a few seconds (I am serious here).

 

One big warning : To get NOS/Filterless real good, it takes MUCH more from your further equipment. This is because transients remain transients, and whatever can't follow will plainly distort.

So yes, I use the most fast amps imagineable, and no I don't use a preamp, and yes my speakers (this is not only tweeters !!) can cope.

Without this all, you'll have smeared and "too fresh" sound only.

 

Another 2c,

Peter

 

 

[audioengr]

I dont believe that bandwidth limit of the high-frequencies is the problem. Its the jitter, noise and distortion layered on top of the music that most audiophiles dont realize is even there. Once removed, you have the violin.

 

It is correct, that the rest of the music system must be up to the task, and that includes minimum compression of dynamics, good HF extension, (not ultrasonic though), and tight control.

 

What I mean by tight control is that when the musical transient occurs, that the speaker is able to react linearly to the voltage stimiulus without overshooting. Most speakers run "open-loop", so you dont really know whether the driver diaphram or other surface is actually moving correctly to reproduce the music waveform. There is also the effect of the imperfect devices in the speaker crossover. For instance, the tweeter always has a series LF limiting capacitor. The quality of this capacitor is critical to prevent other distortions from ocurring, such as dispersion caused by dielectric absorption.

 

A good example of tight control is a powered subwoofer that uses critically damped servo-feedback amplifier. The difference between this and an open-loop sub is night and day usually. you can of course design a bad one, just like anything else....

 

The conclusion is that because things like piano and Cello/ double-Bass and Violin have so much harmonic content and are at the same time percussion-like, they are the most difficult for systems to accurately reproduce. Everything, and I mean everything in the system must be near-perfect and optimal to get a live result from these. I get close I believe because everything in my system is specially selected(some cables), modded or one of my own products.

 

Systems that are not heavily modded by professionals will never get there IMO.

 

Steve N.

Empirical Audio

 

[syganymede]

Sorry all, but the link to the article has been moved as I sorted out my web sites. The new link is on

 

www.cottage-audio.co.uk

 

 

 

 

 

[PeterSt]

The conclusion is that because things like piano and Cello/ double-Bass and Violin have so much harmonic content and are at the same time percussion-like, they are the most difficult for systems to accurately reproduce.

 

It may be known to some that I have been the most surprised how an instrument like a double bass improves from -in my case- a very well controlled PSU section in the DAC. The "danger" with a double bass, and also bass/cello, is that they tend to sound like played softly (especially when the strings are plucked) when things are not right. This is the percussion-like as Steve mentioned. The rather metal sound coming from a hard plucked string just is not there when your system isn't on par, and it just sounds like the player only plucks the string a few mm while it is several cm.

 

Because this sound is pure metal (rather high pitched), it seems odd that such a profound sound can be "forgotten" by systems. Again, the danger is that you will think all is right, and the player just plays the instrument more soft.

 

For a piano/wing another surprise came to me with the better DAC : People are not able to play chords with all the keys being pressed exactly the same time. Or otherwise it's just not done.

An OS DAC can not show this and everbody can press the keys at the same time.

 

It is these things which are very important to the foot tapping aspect, or the getting involved;

Players create the (or their) mood by those very very small variations in playing, and it is these variations which cannot be heard in the not all the best systems.

 

I grew up with a father (first solo player in the orchestra) ever practicing the alt-violin, and I can very well tell when the mood the player has put in, comes through. I don't know of any other instrument where this works out best *and* always is applied (opposed to e.g. an electrical guitar which sure can do it too, but where this is not always applied).

 

An OS DAC does not show moods.

 

Peter

 

PS: Steve, my Duelund 1Kg capacitors of 400 euro each are probably the ones you mean. haha

 

[audioengr]

Peter - you and I are definitely on the same page. When the system is up to it, the passion in the music definitely comes through, with all of the things you mention. When the hairs on the back of your neck stand-up, you know you are on the right track.

 

I know the Duelund guys BTW. They come and see me whenever I'm at CES. Nice guys and great products. For inductors I also like Hans Jensen's at jensencapacitor.com. These Danish guys all know each other.

 

Steve N.

 

[audiozorro]

Plenty of theory here, especially concerning OS and NOS DACs. Regardless, it still boils down to the best implementation that maximizes the virtues and minimizes the shortcomings in either design.

 

The Wavelength DACs are also NOS DACs and perhaps Gordon shares similar views with PeterSt and SteveN. Perhaps SteveN can expound on the differences between the $200 VALAB DAC and his own $3700 Overdrive DAC. At 18 times the price one expects a world of differences.

 

And while $3.7K is expensive for many, it’s not expensive for high end. How about the $69K GTE Trinity OS DAC and how does it compare to Audio Note NOS DAC 5 for a measly $33K? Fortunately we have other more reasonable high end DACs from Berkeley Audio, Esoteric, dCS, Weiss, and others.

 

And obviously it’s more than the DAC chip itself. Otherwise why pay $13K for the Esoteric D-03 DAC with the AD1955 chip, while the Lavry DA10 DAC which uses the same chip costs less than $1K?

 

The audio reviewers are often little help when one review reads very similar to another and the substantial difference is that they sound different as opposed to better. Of course you have the other extreme where the reviewer tells you to hock your car, take out a second mortgage, raid your 401K and run, don’t walk to the nearest dealer.

 

Whatever you believe is the very best sounding DAC today, do you really think there won’t be better sounding less expensive DACs 3 years from now?

 

Finally, I meant to add that I wholly agree with the syganymede original post. While lossy audio or 16bit/44.1kHz audio can sound good, 24/88.2 and higher resolutions almost always sound better.

 

 

 

 

[BEEMB]

 

I'm probably missing the point here, but, if we can't hear anything above 20khz, then why do we need to record or reproduce it? I'm sure that many instruments produce 'something' above 20khz but the majority of us can't hear above 16-18khz by the time we're in our 30's. If those frequencies have the ability to impact audio below 20khz then well, we're hearing that impact since it would have been recorded.

 

Seems a little pointless but I guess it's great, or not so, for our dogs.

 

Matt.

 

 

 

[One and a half]

Apart from studio generated audio, this page lists most instruments including vocals and their frequency range, includes overtones, if any up, to their limit.

 

http://www.independentrecording.net/irn/resources/freqchart/main_display.htm

 

The difficulty is the speaker as Steve described, the speaker is an open ended loop. For most industrial processes, the calculated output is fed back to the process to correct for any errors. This can be high precision, such as placing components on a printed circuit board to within fractions of a millimeter by a robotic arm, or preforming remote surgery.

 

So in a Hi-Fi case, our speakers would need to send a signal back to the amplifier so the amplifier can perceive what the speaker is responding to and make a correction. If you have a speaker that can't resolve, say 18kHz at 0.001db compared to your reference, no manner of juggling gains at that those frequencies is going to correct squat, and the sound will be terrible as it tries to overcompensate.

This is before the listening room starts to play havoc into the equation! Granted some DSP's can very accurately work out what's missing, and compensate, but the speaker transducer is such a long way from being close to "perfection", we still have a long way to go.

 

As for a newer medium for high quality recordings, which one do you choose, You have listed three formats, AIFF, WAV or Flac, would a commercial enterprise store three different versions so the user would pick and choose? Everyone would argue about their preferences and the situation would go nowhere. SA-CD was tried and failed. I agree that better Sound Quality downloads are needed on a larger commercial scale, however with small beginnings such as Blue Coast Records offering WAV at 24/96 is a start and will gain acceptance and creat a niche demand which will I'm sure kick along, now we have the world as a market, not just at the whim of recording companies with tired and dying distribution methods.

 

From WAV you can convert to your favoured format anyway. As for mainstream acceptance of this format when the iPod has limited flash memory and the end output is (usually) a pair of little white ear buds...you going to get this to change? Good luck I say.

 

i just wish Blue Coast would have tracks that aren't so dreary at times! (Just a personal opinion)

 

[blueixus]

I would like to thank PeterSt for the brilliant explanation of the detrimental effect of OS Dacs on the nuances of Music.

 

For me higher sampling rates are just about sound quality. Most agree that good analogue sources sound inherently musical and they don’t have great expanses of high frequency reach.

 

I am nearly 50, so my hearing won’t reach 20Khz, but that does not mean I am less fussy about how my music sounds!

 

My ear is still very fussy about violins, pianos etc and I have to agree about the naturalness of NOS Dacs. I regularly use a NOS Audionote, I have recently been playing with a NOS DDDac kit based on a TD1543, this has no right to sound as shamelessly musical as it does – it actually quite shocked me!

 

And the point that NOS Dacs place a burden on the rest of the system is absolutely right. It has taken a long time to get my system right within this context and the speaker crossover was a major part of this, active systems to one side, a lot of the musicality is lost through lazy design and a penny pinching choice of components.

 

Top flight stuff from the likes of Dueland, Jensen and Mundorf can make huge differences in this area.

 

 

[audiozorro]

The frequency ranges of many instruments go beyond 20kHz. In the common frequency range charts the data shown for instruments refer to fundamentals, and do not take harmonic overtones into account. The link below is for research done at the California Institute of Technology that demonstrates the musical content of instruments above 20kHz.

 

http://www.cco.caltech.edu/~boyk/spectra/spectra.htm

 

I have posted other links in previous CA threads that show the capabilities of human hearing. Another one is posted below for your pleasure.

 

http://www.silcom.com/~aludwig/EARS.htm

 

Please remember that two ear hearing test results are much different than one ear hearing tests.

 

Finally, on the matter of the common audio file formats, AIFF, FLAC, and WAV, the burden should be on the player software vendor to accommodate and not on the end user to have to choose or convert. Two of the players I use most often, MediaMonkey and XXHighEnd understand and provide this flexibility. Another player I use, iTunes does not, though Amarra promises a future version of their software will provide FLAC capability to the standard AIFF and WAV formats in iTunes.

 

 

 

[audioengr]

"Perhaps SteveN can expound on the differences between the $200 VALAB DAC and his own $3700 Overdrive DAC. At 18 times the price one expects a world of differences."

 

In addition to sounding a "world" better, it has these features:

 

1) balanced outputs

2) ultra-low noise volume control - just like connecting line-outs direct to amps

3) enough drive for Sennheiser headphones

4) two-chassis power system enabling battery option

5) muting for power on and power off - no clicks to damage tweeters

6) not NOS, but selectable digital filters allow it to sound like NOS

7) two gain settings

8) adjustable volume range

9) deemphasis switch - great for old 70's tracks like Hendrix and Led Zep

 

The bass from it is unlike any DAC on the market. The clarity, particularly when driving amps directly is unparalleled. More live than any DAC I have heard, and I've heard most of them. I dont expect any of you to believe what I say, so just read the feedbacks from my customers that have them:

 

http://www.audiocircle.com/circles/index.php?topic=65385.0

 

Steve N.

 

[audioengr]

"The audio reviewers are often little help when one review reads very similar to another and the substantial difference is that they sound different as opposed to better."

 

This is what makes this hobby so difficult. Reviewers NEVER do shootouts, or rarely. They dont want to jeopardize future AD sales or step on toes. They have to walk the thin line. The only real shootout I'm aware of is:

http://www.positive-feedback.com/Issue39/ramblings_computer.htm

 

There was a DAC shootout at BAAS in the last 6 months. You can read about it here if you search:

http://www.baasnotes.com

 

I just did a demo of the Overdrive for the BAAS three weeks ago.

 

And I forgot to add one more feature - unlike most NOS DACs, the Overdrive supports 24/192, 24/96 on the USB input.

 

Steve N.

 

[BEEMB]

 

Okay, so to summarise then ... there are benefits to a system being able to reproduce higher frequencies because it could possibly affect lower frequencies which we can hear.

 

Is that what we're saying.?

 

 

 

 

[BobH]

Here's a little light bed-time reading.

 

http://www.cco.caltech.edu/~boyk/spectra/spectra.htm

 

Goes a long way to making it less easy to disagree! I reckon it's one of those odd things that is never satisfactorily explained because no-one really knows the answer! I much prefer a NOS Dac to an OS one but there are equal arguments and reviews in favour of the OS approach. I've yet to hear anything oversampled that I preferred to the original, either doing it myself or if someone else has done it for me.

 

Right OT, the Society of Sound, from B&W, have recently made their back catalogue available to members and some of these have been upsampled from the originals using Pro Tools Tweakhead software. In direct comparison between the original 16/44.1 and the software upsampled 24/48 I could hear the difference and much preferred the original. Where the tracks had been downsampled to 16/44.1 (for the original release) and I was listening to the original 24/48, I preferred the original! So, for me, not messing about with the original signal seems to be preferrable to an outright 'numbers' chase. In this I think Peter has it right on the button, if I understand him correctly. It's the messing about that causes the problems, not the resolution per se.

 

[audioengr]

Upsampling is something I have a lot of experience with, and have learned a lot. Having modded 15 DACs from different companies and tried several S/W upsamplers, both PC and Mac, here are my conclusions:

 

1) Hardware upsamplers are generally inferior to software upsamplers

2) On-the-fly software upsampling code (like SRC) is inferior to batch upsampling codes (overnight type stuff like R8Brain or Adobe Audition)

3) The sample rate and the word length make a HUGE difference with each D/A chip, particularly NOS DACs - for instance, the VALAB DAC, which is an old Philips 1543, sounds best at 20-bit and 48kHz, even though its a 16-bit chip.

 

I personally like 24/96 upsampling with either SRC and Foobar 0.8.3 or iTunes version 8. I have found that these sound very good on every DAC I have tried.

 

All of the above observations make me very suspicious of anyone trying to draw general conclusions about upsampling. The answer IMO, is "it depends".

 

Steve N.

Empirical Audio

 

[PeterSt]

The topic resurrects at many places, but what you must try to get is that an oversampling DAC is not the same as an upsampling DAC (keep in mind though that oversampling/upsapling is not really defined, so I use my own definitions for something which is really different in practice).

 

An upsampling DAC may take 44.1 for input, and outputs that at 176.4. Or takes 96 for input, and outputs at 192.

This would be the good case, while most upsampling DACs don't make use of this good case, and instead they upsample standard to 192, no matter the input rate.

 

Anyway, this is what Bob described as "I never like that", and it would be similar to upsample in software, of which Steve says "I never like the real time solutions".

 

But this is only upsampling, and although *I* never like that either (and I really tried), this is not what I am talking about.

 

First hear this : any DAC you'll meet doing 24/192KHz (mind the 24 !) is an oversampling DAC (unless you meet my own haha). Oversampling may be described as the process necessary to let the DAC work in the first place, because at least a part of it is single bit operating.

 

A pure sigma-delta DAC is single bit, and you could say that the oversampling principle allows the "creation" of more bits. Keep in mind though, that we are talking about 256 times "upsamping" internally, followed by a virtual other way around by means of filtering.

 

Many variants of sigma-delta DACs exist, e.g. using the 5 most siginificant bits as ladder (R2R) and the remainder treated as single bit again. In either case it needs oversampling to operate.

 

When you don't like upsampling from 44.1 to e.g. 88.2, it will be a tough job to like oversampling from 44.1 to 11289.6 (all has become sines now).

 

Keep in mind that once you're into real oversampling, the process of what your are listening to has changed. I mean, at going from 44.1 to 88.2 it is only a matter of more or less decent interpolation, which never is reality. A negative. The *reason* you'd do this, is to shift Nyquist, and turn some squary waves into more better sines, *if* they were sines in the first place. If not, this would just do wrong (go figure what is better -> undoable, BUT it is audibly different).

 

Going from 44.1 to 11289.6 is not done for te reason of shifting Nyquist, and while it is the means to let the DAC operate, it needs a filter to get rid of high frequency noise. Without this filter, no sound. Here another process of destorying takes place.

 

The mildly upsampling DAC doesn't need a filter at all, BUT you wil have piles of harmonic destortion.

The not upsampling DAC (so, not OS and not US) also doesn't need a filter at all, BUT the pile of harmonic destortion would even be larger.

And now the big trick : An upsampling DAC (without oversampling !) to 352800, does not need a filter at all, because all (or enough) Nyquist related anomalies have been shifted out of the audio band.

 

If you still can follow (sorry for my english !), you'd now understand that upsampling in software (in real time or off line) is without any sense when an oversampling DAC is used. You'd only avoid one step out of the 256 at doing it. It makes no sense, and probably it makes it worse because "your" upsampling won't be consistently working with the DACs oversampling (which is just a step of "upsampling" by itself, but with a filter at the end).

 

One other thing :

As some may know I kick against official measuring which will tell that OS DACs may contain 0.0002% THD+N (total harmonic distortion including noise). While I already could reason this is total BS (if measuring manufacturers are offended, so be it ... :-), I can now clearly see it by my own measuring.

It is the most clear to me now, that whatever goes into the DAC by no means comes out of it. We may even be glad to perceive some music from it. In a later stadium I will show graphs of it, which right now still is a difficult task because it needs the direct comparison from a source file with the analogue output. This is a most difficult project, because it needs proper alignment and many tricks to get there. But I'm sure I will. I only wanted to say :

 

The output of a DAC is so much off compared to its input, that most probably the harmonic distortion as incurred for by NOS/Filterless it near totally flattened, because harmonic distortions comprise of peaks (though with slopes but they are way below normal wave aplitude output), and there is no way these peaks will come out. This by itself is because there is no way the DAC as a whole (incl. I/V stage) can follow this, and as I said elsewhere, never mind the slew rate and settle times the datasheets talk about.

 

For a similar reason the OS DAC making sines from e.g. nice squares from a synthesizer, those squares are already flattened by the NOS DAC just the same. So, the OS and the NOS are more close to eachother opposed to what we may think looking at the digital wave data, thinking that the NOS DAC would output that near 1:1.

No way !

 

The very main crux here is : just because the difference between the input and the output is so huge, there is so so much to do about the NOS outputting more 1:1. A change of a capacitor may give a change of day and night already.

 

On this matter, my own super shunt I/V regulated DAC, now is being replaced with a FAST super shunt regulated I/V stage. So, super shunt is one, but fast super shut just is another. What will come from this ?

 

... bumping into limits probably. The harmonic distortion will be more profound, and maybe the NOS DAC will become unlistenable because of it. Already by itself, but also because speakers may not be able to follow the transients anymore.

We'll see.

 

Peter

 

[BEEMB]

 

Bob,

 

I like to make a statement and get a load of replies.

 

I have to say I'm with you totally on the "don't mess with it" front. Output the original data as is, no messing, untouched. Whether I can hear that or not depends on whether I'm blindfolded (can't tell) or not (can tell).

 

My brother and I will get around to some more testing soon.

 

Matt.

 

[BobH]

Well, you certainly got a load of reply!

 

Sounds like you and I, at least, are on the same page when it comes to all these 'differences'. Those that I know are there are much easier to detect than those I don't know about!

 

Mind you, some of 'em show up. I was listening last night to the Blue Coast Collection and, even though it was the first time I'd listened to it, somehow it didn't sound as good as I was expecting it to. Quite by chance I pressed the volume button on my Squeezebox remote and the volume changed. Normally this would be a good thing, except I'd set it to not do that! A quick trip to the settings page and it turns out that the gremlins had set the volume to be active. Set it back to inactive and all sprang quickly back to life! So at least I'm confident that, on my system, I can detect when it isn't how it should be, volume-wise.

 

Over extended listening I 'prefer' not to have any over/under/sideways/back-to-front sampling but it is not an instant 'that's horrible' kind of thing. It is more an 'overall - but can't quite put my finger on it' sort of thing.

 

When I've heard them, that has also been my experience with Super Tweeters. Can't hear a difference until they're not there. And the same goes for hi-res audio. I don't really notice a difference until I go back to the original - then it's obvious. But a well recorded anything beats a badly recorded anything, every time.

 

[audioengr]

Peter - I agree with you 100%. Oversampling, and particularly digital filtering is the bain of digital audio. It's what makes it sound not like analog, along with the jitter of course. These two things are the main obstacles for digital audio IMO.

 

This is why these older chips with no digital filtering are gaining so much popularity. Unfortunately you usually sacrifice resolution because they are old and dont deliver the benefits of 24/96 or 24/192 etc..

 

Steve N.

Empirical Audio