Rupert told me to use mic amps with very extended freq. response. Thus assuring proper reconstruction of the delicate overtones from acoustic instruments.

These include trumpets, organs, recorders, perc., etc.

Therefore here I use the following:

Rens Heijnis Custom Built preamps---3dB @ 1 MHz.
Crookwood Paintpot -3dB@ 500 kHz.
Stagetec Nexus Compact -3dB@ 700 kHz
Rupert Neve Designs 5012 -3dB@ 200 kHz
Sonosax -3dB@ 300kHz, true 90 kHz with 192 speed recording

Theory men, Princes, and Euro opiners / naysayers step off.

The recorded result is CLEAR.

Analog systems are very different from digital ones, though. I think this is where you still mostly get hung up.
There is a very good reason for wider bandwidth and gently sloping filters in analog circuitry. You need that for accurate (linear) phase response. Amplitude and phase are correlated in analog systems.

Digital filters don't have the same requirements because amplitude and phase response are not necessarily connected (although they can be in IIR filters). You can have linear phase right up to the brickwall cut off frequency with a FIR filter (which is what well designed converters use).
This is why with low sample rates you can have the best of both worlds (in terms of accuracy): correct amplitude/phase response AND limited bandwidth for lower IMD.

There is no such thing as the more the better in real life. There always is a range, above and below which performance starts to decrease. With audio that range is very narrow. You need at least 35 kHz sample rate to cover up to ~16/17 kHz, past which most people can't hear. You want to account for the 0.00....01% of the people who can hear up to 20 (22?) Khz.
Anything above that is overkill and starts to be detrimental. Luckily our hearing is low revealing enough that we can't even tell most of the time, so you'll be safe using 192 kHz or above.

But there is a difference. No question. At least a measurable one. You seem to (or claim to) be able to always tell that difference, and you like the additional IMD provided by ultrasonics (always? That's a little weird but who am I to question your taste?). That's all fine, but it's not an increase in accuracy that you're hearing.

All you guys needed to do the day of that test was put a measurement mic in front of the speakers and you would have seen clear as day the reason for the difference in sound: IMD components that are not supposed to be there (if one is after maximum accuracy. Which they don't necessarily have to be, of course).

If someone still doesn't believe this, do the same Neve test but this time using speakers with a dedicated supertweeter that is the only driver that reproduces the ultrasonic sinewave, while the other audible sinewave is reproduced by the regular tweeter. If each driver produces only one sinewave, you should hear no difference... still, make sure to use a measurement mic in case some IMD gremlin still makes its way down into the audible band. Speakers really are "special" equipment specimens.

Seems to happen ALOT around here. There are a lot of miserable know it all's that don't take responsibility for their negativity. Not talking about anybody in particular.

Are you saying that Monty's video proves that ADDA converters can reproduce the analog input perfectly, hence the OP should be able to produce perfect digital copies of his tapes that sound the same as his tapes?

For some people, “prick” is a synonym for logic and fact

Lotta argument from authority goin' on here.

They can reproduce a band-limited analogue input pretty perfectly. Yes. A tape machine signal is not band-limited and therefore there will be an audible difference if your sound system is prone to IMD or other sources of non-linearity.

A tape recording that's gone through a high-end AD/DA loop at a high enough sampling rate (say, 384 kHz) to capture the tape bias signal will sound exactly like the tape recording, yes. It will also measure exactly like the tape recording.

Whether you think that tape bias is a valuable thing to capture it's a matter of debate – maybe. It certainly changes the way audio equipment responds to the input, and probably audibly. I'm sure having a constant sine wave up there also changes the response of things like compressors, too.

A brickwall low-pass filter, whether in analogue or digital, will change the signal somehow. The audibility is debatable, but comparing to tape is very different than comparing to an acoustic instrument signal, because usually you don't have all the unnatural ultrasonic weirdness tape has in a real-world acoustic performance!

If you want to capture a signal faithfully, there is absolutely zero doubt that analogue tape does a worse job. It's super noisy, requires ultrasonic bias to have a decent frequency response which will probably cause IMD somewhere along the way, has measurable wow and flutter, print-through, hysteresis, dynamic range compression, and old-fashioned saturation. All of which are absent in digital audio.

The only changes in audibility in high-quality digital audio are the phase response around Nyquist if the filter needs to be low latency (this is not a limitation for non-real-time downsampling btw), and the fact there is nothing above Nyquist. Quantisation distortion is zero if the signal has a level of noise above –138 dBFS (to be very conservative), which no converter has. Aliasing is literally not a thing for any high frequency sample rate, and arguably not even at 44.1k in 2025. But again, if you want to capture the weird, highly unnatural signals that exist in analogue tape above 20k, you absolutely can in digital too – you just need to go to a much higher sample rate.

Now, I'm not gonna lie, tape sounds very nice. But it sounds far less like the original input than digital audio does.

Going back to the question, yes, Monty proves that the signal is faithfully captured and reproduced. A signal with nothing above Nyquist, which tape isn't at 44.1 kHz.

In theory you're right, in practice you're wrong.
Find me any chip manufacturer that claims perfect conversion.

Pretty wacky. Every tape machine is absolutely band limited, always has been.

Yeah, well, you can’t even do analog tape recording without bias, it doesn’t work. It might be a good idea to gain understanding of the systems before criticizing them.

No. This is not always true at all.

You can’t ever actually compare a real world performance to a recording. They are entirely different acoustic events.

Again, only a smattering of understanding here. Bias doesn’t cause IMD, it reduces it.

And I am in no way, defending analog tape!

Probably should also study up on the audibility of phase shift. It’s not what you might expect.

The presence of signals above 20kHz on tape is not evidence of its ability to capture ultrasonics. The tape system is laden with nonlinear elements that generate harmonic and intermodulation distortion. The 3rd harmonic of 8kHz is 24khz. Yet the total system response falls rapidly above 20 kHz. There will always be signals generated by the nonlinearities built in that aren’t in the original.

Yup.

Tape response about 20Khz isn’t as different from 44.1 digits as many think. It’s mostly not capturing ultrasonics either, but it definitely will generate them through distortion.

Not through a sharp brickwall low-pass filter like digital is. Good tape can often record up to 40 kHz with a good condenser mic. This is pretty obvious if you record a good tape recording at a high sample rate and look at the spectrograph – there will be a noticeable amount of ultrasonic content. You can actually pitch it down and hear it, and it's not (just) distortion harmonics – it's part of the acoustic signal. Also, as I mentioned, tape has an ultrasonic bias sine wave playing back. That's emphatically different to the absolute silence above Nyquist a digital signal has.
Just to clarify, I don't believe this ultrasonic content is useful or good. I'm just saying, some playback systems might distort because of it in an audible way.
Of course you can record without bias. It just distorts more and has a poor frequency response.
To specify, there is always a change if there is anything above Nyquist. And I'm not saying that it's audible. I'm just saying, the signal is literally not identical, because the whole point of the filter is to change the signal (in a useful way). I probably should have been clearer but this is just semantics.
Of course you can. And you should. Because it's a great sanity check. If your speaker and mic are good enough, a close-mic recording of something like a violin through a great speaker put in the same position as the violin player sounds surprisingly life-like. More so through digital than through tape, too.
Bias reduces IMD within the tape recording itself. That's kind of obvious and not what I meant.

From my understanding, the bias signal is not filtered before the audio signal is output from the tape machine. Which means there's a big ole sine wave in the ultrasonic range during playback of a tape going into the rest of your equipment. Some of that equipment might not behave linearly with it there.

I might be wrong and tape machines might have a low-pass filter to attenuate bias during playback. Again, my understanding is that it doesn't.
I have actually done several tests about this, and my conclusion is that I don't hear the phase shift of a sharp minimum-phase filter at 22k if the treble tonality isn't altered. But I do not deny the possibility of it being perceptible in a poor implementation, seeing as it does change the signal within the audible range.
I completely agree that distortion adds harmonics in the ultrasonic range, but I'm not sure where you get the idea that a tape machine cannot record things above 20 kHz. Not only can many microphones send those frequencies to the tape, but the tape literally records bias, which is an ultrasonic signal. There is no filtering there. Whether the ultrasonic signal comes from a microphone or the tape bias oscillator is literally unimportant to the fact it's in fact recorded by the magnetic tape.
From the above nitpicking it might sound like I disagree with this, but I don't. Because ultrasonics from acoustic instruments aren't anything magical or fancy. It's just a bit of airiness, and your ears don't have a mechanism to pick them up anyway.

Why would 44.1 and 48 kHz sound much worse than higher sample rates?

I would say that, given a competently designed converter. Given that 16/44.1 already exceeds the practical limits of human audibility, we would need a good blind test to prove otherwise.

An ABX test with a tape machine and an ADDA loop would be a good one. I believe that has been done in the past but it might have been with a phonograph instead of a tape machine.

15 year old thread.


sheesh.

Your grasp of the formation of Latin plurals seems to be on point.

But in terms of contemporary technical usage, particularly in mathematics, theorem has a distinct meaning. Feel free to look it up. I already did.

Yes, you mentioned that a few pages ago. And yes, I'm aware of the usage of theorem vs. theory in mathematics and science, and I'll also admit that I was being cute... attempting to poke some fun into a contentious thread that, in retrospect, would be difficult to interpret as "fun" with the cold delivery of internet text! So, I apologize for that.

I think it's also, however, important in the current age to emphasize how strong a theory has to be in order to remain theory in scientific consensus. The "it's just a theory!" crowd that likes to deny all kinds of scientific truths doesn't understand that, in science, theory – much like a mathematical theorem – has held up to scrutiny and remains strong. It's all too common for folks to confuse "theory" for "hypothesis."

Either way, I was honestly agreeing with you! Just being a bit silly about it.

This drives me crazy too.

early Ampex machines had no intentional low pass filter (bias trap).
later units especially those with Sel-Sync were fitted.
MM-1000 had 2 sets of traps, one in the electronics assembly and another in the front Sel-Sync panel.
In each case, the filter only affected sync, not repro.
Scully 280 placed their bias trap on the main output, affecting repro, sync, and input.

Thank you, that's really interesting. I'd love to see the phase response on that notch filter in a graph

It is my experience and understanding that all the converters sound different, and that this is a fact of physics which any actual engineer will confirm, including those who built them. To say the Scarlett sounds just like the Lavry seems a lack of recognition of the effort and ingenuity which went into it.

The cheap converters certainly sound miraculous these days, and there are no issues recording professional albums with virtually any one of them. However, there seems to be a noted improvement in sound quality, commensurate to the price, as you move up to something like the RME ADI Pro 2, or the Aurora, or the Lavry.

When you make a claim like the Scarlett measures to be identical to the Lavry, I really question your measurements and methodology, or the usefulness of such measurements at all. There are no simplistic blanket statements in this game. If it were that easy, we'd all be famous producers and engineers.

There's no simplistic explanation given in these pages why there could be an audible difference. That doesn't mean we can't make the correct statement that properly implemented 44.1 kHz is cleaner than higher sample rates once it comes out of your speakers. That is a fact, not a simplistic statement.
What keeps people from becoming famous producers and engineers is their artistic skillset, not the converters they use, or the (lack of) understanding of what goes on inside them.

The builders obviously need to know what goes on inside, and it helps if they have some artistic sense too. But they do not have all the answers. This idea that science or measurements have all the answers when it comes to audio or music is psychological and has nothing to do with building gear, or making music. I have worked with a few legendary builders, and they did not have all the answers or explanations of things, and never speak in absolutes.

Now you are making the absolute claim that 44.1kHz, properly implemented, is "always" "cleaner" than higher sample rates. It may be true that higher sample rates always introduce some audible noise at the extremes. But I am unsure as to the constructive purpose of these discussions except to prove who is right, who is the expert.

And the claims made here are often unclear. It seems the going wisdom here is that all the converters are an identical, perfect conversion due to Nyquist. Even if this is so, which is highly questionable due to its limitations in practice, but at least the analog components are also affecting the sound, are they not?

Those who have built gear, or produced music, quite extensively, are those who know the answer to these questions, and the extent of their knowledge is the extent of their success in the industry. Those who sit on the sidelines, speculating about it, watching videos or reading about it, the extent of their knowledge on it is approaching zero.

There is no great lab coat scientist in the sky who has done all the scientific tests on everything in audio, music, and life, has all the answers and the absolute scientific proof of everything.

Nobody is claiming that all converters sound the same. We discussed some of the reasons why they may sound different. Sighted tests being the main one, by the way. Some people even argue that there is no proper test if not sighted, as in real life you always know what piece of gear you're using.

While the philosophical implications of such a claim are interesting to ponder, it is in no way a scientific way to investigate this matter. But if a piece of gear makes you feel good, or a sample rate makes you feel good, by all means you should go for it. If nothing else it should free up your mind and let you get more creative.

However, if you want to try to step out of the subjectivist realm (it gets lonely in there sometimes), look at extensive measurement sets of converters like ASR does (and I am no fan of ASR, if not for just their electronic device measurements). Or Archimago (a much more knowledgeable guy).

Now compare those measurements with speakers measurements (which are even worse after you put them in your room to listen to them). Erin's audio corner is probably the better source for this, but then again it all pretty much comes crashing down once you're not in anechoic conditions anymore.

Pay attention to the speakers distortion/compression graphs and draw the logical consequences.

I wouldn't be making the same absolute claims if we were talking about speakers, because test results of those are not absolute once you take them out of the anechoic chamber. Indicative, sure. But not absolute by any stretch of the imagination. Also, if we really want to be sticklers, for speakers a truly complete set of measurements is impossible to get. You would need infinite measurements for directivity, for example. Still, the spinorama provided by Erin (or ASR) is good enough for all practical applications.

For converters it's a different story. Unless we're talking about a converter with unusual input/output impedance values, preceeded or followed by unusual impedance gear, what you test on the bench is what you get in your room. It's just voltage on a wire you're measuring. It is not an uncharted territory.

In all of this the most interesting aspect to me is that to get complete sets of measurements one has to rely on third parties. You can make of this what you want, but to me it doesn't shine a good light on the audio gear industry as a whole, especially gear that comes with hyperbolic claims attached.

44.1 cleaner? It is an absurd contention and proven wrong in any playback room of high quality.

I challenge your hearing.

You are not a practitioner. You are a pot stirrer and antagonist. Pro antagonist on GS.

You Prince

You can go here yourself and take a look at professional measurements of speakers.

https://www.erinsaudiocorner.com/loudspeakers/

-30dB harmonic distortion (3%, any order) is common. -40dB (1%) is considered very good and rarely seen. This describes a non linear device. Arguably the most non linear device in anybody's audio chain (save for the intentionally non linear ones such as compressors and limiters).

I think I already explained well enough what that means in terms of distortion components passed down from the ultrasonic region into the audio band, by now.

And many common nearfields that try to turn out to be distortion machines trying for volume and extension from small sizes and cheap parts.
Don’t measure the long throw woofers trying to play sub and 200 hz at the same time! The small three ways are the biggest offenders.
Don’t measure the small subs in any barefoot type design or anything that looks like a smaller barefoot.
Don’t measure woofers of the Genelecs ones and Neumann kh and related designs that boost the small woofers a great deal in the crossover and limit them! You’ll measure or hear the soft clipper for transients! It doesn’t matter if you paid thousands.
Don’t measure ribbon tweeters and AMTs. They are usually much higher distortion than domes.
Don’t measure power compression in the ones that use generic drivers. The vertically integrated manufacturers have a huge leg up with their upper end designs here and what they’re able to afford to put in a pair of monitors that have to cost a few thousand dollars retail.

PMC monitors get around these problems with a transmission line design.

A serious PMC system starts at $20,000.

THIS is Prince:

You are showing admirable restraint.

while its entirely subjective, my 35 years of experience with analog and digital is far more important to me than any "test". the obsession with tests and graphs and charts is always overwhelming to me.

48 hours ago, I brought one of of two A827's online. it has not yet been calibrated. it is off. that is immediately apparent (the tech insists on de-magging before I put the MRL up and do a proper set up - he can't come until next week)

that being said, running a CD of Linton Kwesi Johnson's first record on CD through a Burl B2 then the Manley pultec and then two channels of the A827 w/ the rarely used 24 track headblock.

it sounded better than the CD. immediately. even with wacky meter readings. maybe its the pultec. maybe the transformers in the Burl

but I like it. people want to work with me because I focus on sounds I like.

analog always sounds better digital to my ears.
i am never emotionally or psychologically or spiritually moved in any way by the notion that gods of accuracy and science are frowning on me.

that being said, I have no concern to convince someone else - they may well be having the same emotional experience with their digital

Be well

- Jack

No, they don't. A transmission line cabinet doesn't help when it's the drivers themselves that have this distortion embedded in them.

See here, for example.

https://www.soundstagenetwork.com/in...ents&Itemid=18

And here is how one of the arguably most linear drivers measures. Purifi PTT6.5M



You want to see something really scary? This is how it measures with a 80 Hz tone and 200 to 6000 Hz tones at the same time.



Still.. we don't hear that over music. Luckily our hearing is not that revealing, or we could only stand to listen to live acoustic music.