Digital Audio and Sampling Rates

[UnderTow]

Originally Posted by ARIEL

With his testing methods , Both being mixed inside the box , All plug ins and settings would be identical . I like the fact he did a proper test of this . A multitrack recording at both sample rates and not just a stereo test .

I don't mean any disrespect to Eric Valentine but we do not know if he did proper testing. There simply isn't enough information to ascertain that. What does "mixed in Pro Tools" mean? I have read quite a few stories where people literally mix the same material twice at different rates and think they can draw conclusions from that. I assume that isn't what was done here but we can't be sure.

Also there are plugins that scale their values based on sample rates. So for instance a low-pass filter might end up higher when loading a preset at a different sample rate. That doesn't mean that the user can't set the filter at the same frequency. It just means that the internal values will be different for the same cut-off frequency at different sample rates. This kind of difference will have no bearing on how someone mixes when they adjust the cut-off by ear but it will completely invalidate a test in which presets are copied from one sample rate to another.

Even if the tests were all performed to the highest possible standards and all the listening tests double blind, it still doesn't really tell us anything new. We already know that some converters sound better at higher sample rates. That is a given but one can not generalise from that that all products sound better at higher rates. That information about a particular interface is useful if you own that product or are in the market for buying a new interface but otherwise as a general rule to apply to all interfaces, it isn't really that useful.

Unfortunately the article doesn't mention which interface was used. (Unless I missed it). I am guessing the 192 I/O which is a discontinued product. So maybe an interface that isn't being manufactured any more performs less well at the base rate. That is a possible conclusion but we can not conclude that all converters sound worse at the base rate.

Also we know that some plugins sound better at higher sample rates. That isn't news either but again we do not know which plugins were used in the test. All it takes is a single rogue plugin that doesn't behave well at the base rate to skew the whole test. In my opinion in means that such a plugin is best avoided rather than squandering resources on running everything at higher sample rates to make up for the deficiencies of a single tool. For others such a plugin might be crucial to their workflow or mixing style and thus worth increasing the sample rate. Or for some it might just be easier to run everything at the higher rate and never have to worry about any tools performing less well at the base rate.

Of course if you have the resources available then go ahead and run everything at higher rates but IMO it would be more productive and better in the long term for us as customers if we held the manufacturers and software developers to a higher standard. IMO, audio tools should work perfectly well at the base sample rate.

Alistair

[Trakworx]

Originally Posted by UnderTow

Also we know that some plugins sound better at higher sample rates. That isn't news either but again we do not know what plugins were used in the test. All it takes is a single rogue plugin that doesn't behave well at the base rate to skew the whole test. In my opinion in means that such a plugin is best avoided rather than squandering resources on running everything at higher sample rates to make up for the deficiencies of a single tool. For others such a plugin might be crucial to their workflow or mixing style and thus worth increasing the sample rate. Or for some it might just be easier to run everything at the higher rate and never have to worry about any tools performing less well at the base rate.

Thank you for addressing plug-ins!

It seems to me that knowing which plug-ins do and which do not sound better at higher SRs is crucial in making the decision of which SR to work at, or which plug-ins to avoid.

This is information that I have not been able to find, and admittedly I have not taken the time to test my collection of plug-ins. That would be a huge PITA.

Anyone know where to find info on which plug-ins perform differently at different SRs?

[raal]

great thread.

when we take the Avid I/Os for a spin we'll do listening tests vs. Auroara 16s with a live drummer and bass @ both 96k and 48k, no plugs. will report back.

[UnderTow]

I added a fourth post in the series. This time about dither and bit depth. Enjoy!

(Click here to go straight to it: Digital Audio and Sampling Rates )

Alistair

[MusicManic]

I don't understand most of the technical info in this thread, but I thank you for what I think you're implying, which is that I don't need to switch to 96k and that I can record at 44.1k and not worry about compromising anything as long as my A/D converter and plugins are quality?
Any reason to go from 44.1k to 48k?
Thanks!

[losvlados]

Thanks a lot for sharing this information.Great post!

[doom64]

The Most Interesting Man In The World doesn't always record audio, but when he does he prefers higher sample rates than 44.1 kHz. And if he does than that's good enough for me.

[peoriashows]

Originally Posted by UnderTow

=========================== Oversampling Converters and Anti-alias filters ==============================================

The flagship converters of the top four ADC chip manufacturers (Cirrus Logic CS5381, Wolfson WM8786, Asahi Kasei ak5394a, TI Burr Brown PCM4222) all run their modulators (the part that actually does the sampling by measuring the analogue voltages), at 5.6448 Mhz for sampling rates of 44.1, 88.2 or 176.4 Khz and 6.1444 Mhz for sampling rates of 48, 96 or 192 Khz. This means that the analogue anti-aliasing filter now only needs to remove frequencies at or above 2.8224 Mhz (or 3,072 Mhz for 48, 96 or 192 Khz sampling rates). Such an analogue filter can be made absolutely transparent at 20Khz (and even higher).

This is great info! Would you be so kind as to confirm something for me if you very familiar with these ADCs. I purchased a used pre-amp in which someone tampered with the Asahi Kasei ADC (originally an AK5394A) and replaced it with a Burr Brown PCM4202 which doesn't appear to have the same pinout, based on reading the data sheets. I wanted another opinion before I confirm this Bozo's digital malpractice.

thanks in advance

[Syncamorea]

Originally Posted by doom64

The Most Interesting Man In The World doesn't always record audio

Actually, the beautiful young engineers he hires do all that messy recording stuff. He's too busy staying thirsty.

[Jazznfunk]

So, the only way to "see" DAC's reconstruction filter is analog oscilloscope?

[UnderTow]

Originally Posted by Jazznfunk

So, the only way to "see" DAC's reconstruction filter is analog oscilloscope?

What do you mean by "seeing the DAC's reconstruction filter" ?

Alistair

[Herb]

Originally Posted by UnderTow

What do you mean by "seeing the DAC's reconstruction filter" ?

Alistair

I would imagine he means seeing the continuous signal produced at the DAC's output.

[Jazznfunk]

Yes, exactly. To visually compare iZotope RX / Audition reconstructed and real sound card's output waveforms of same test tone files @ various Hz and SR's too.

[UnderTow]

Originally Posted by Jazznfunk

Yes, exactly. To visually compare iZotope RX / Audition reconstructed and real sound card's output waveforms of same test tone files @ various Hz and SR's too.

An analogue oscilloscope would get you there, yes.

Alternatively, you could re-sample the signal with another converter. For instance if you believe the sample rate makes a difference, you could re-sample the output of a 44.1Khz system into a 384 Khz system (Or whatever) and look at the waveforms in the digital domain. That should clearly show that there were no "44.1Khz steps" in the 44.1Khz system. It all really depends on what you would find a satisfactory demonstration of the principles.

Alistair

[string6theory]

This is an excellent thread UnderTow and I don't mean to distract from it. I use PCM regularly and couldn't live without it. It's simply indispensable, and the audio editing/production enabled with 24bit audio is cherished. I personally use 2 Apogee Rosetta 200's and PT.

I also use 2 Korg MR2000S units for mixdown, (some) tracking, and even live 2-track tape machine-style recording, bouncing to the 2nd unit with overdubs, back to the 1st unit, etc. So, I do have lot's of experience with both PCM and DSD - and love both!

Some of your exchanges with the poster Phase Shift regarding the Korg DSD stood out (for me) and I thought I would add my perspective/thoughts if it's ok. Again, I use both PCM and DSD regularly, in well-implemented systems.

I'm sorry if any of your comments below appear out of context as that's not my intent.

==========================================
Originally Posted by UnderTow;

"I don't know exactly how the Korg's converters are implemented..."

"I am not saying the Korg unit doesn't sound good but to my ears it didn't sound any better than the PCM converters I heard in the same studio..."

"...Turning off expectation bias is as impossible as switching off your hearing. (And I don't mean by sticking your fingers in your ears). We can not do it."

"To me this is a marketing product. Not an engineering product. (The same goes for all DSD/SACD products)."

==========================================


I would submit that you may also be subject to an expectation bias in this context. This is not a challenge to what you are "hearing" as it's personally subjective. But, the last comment about the Korg being a marketing product, not an engineering product, is IMHO, just silly. It's seems to bely a sort of dismissive tone that I think is ultimately and intentionally misleading to folks. Maybe you could re-evaluate this?

I've read many, many "dismissive" DSD comments in the past from folks very ingrained in the PCM "industry" as it were (i.e. those making a living designing PCM AD/DA hardware, coding DAW software, designing /selling DAW plug-ins, performing consulting services to inter-PCM industry corporations, etc.) And, as you know, it is a massive industry, with lots of folks seeking their piece of the pie and making decent livings, from trade reporting to association lecturing, to writing industry "papers", to teaching to designing to producing to shilling to... etc.


I'd paraphrase and summarize these "dismissive" DSD comments as follows;

- This technology has been around for decades and they're just cashing in on a patent. (Similar to - it's just a marketing product).
- They use massive filtering, shifting too much "noise" into the ultra sonic frequency range that will build up over multiple tracks, and, this may potentially harm your play back system (or your ears).
- It's only 1-bit and therefore useless in any practical application where audio editing is required, so there's no point using it at all.
- Converting DSD to PCM for editing or final delivery will degrade the audio more than just recording PCM in the first place.
- Multi-bit PCM is the industry standard and ultimately sounds better than 1-bit DSD even at the highest sample rates (5.6MHz).
- PCM converters already have up-sampling to DSD sample rates (before decimation to multi-bit, etc.), so we already have the best of both worlds with this standard.


While there's some truth in each of these comments, I believe most of the conclusions implied or drawn from them are (intentionally?) misleading, incorrect or plain wrong.


My view is that Korg must have put quite a lot of engineering resources into these MR unit's and their implementation. Why? Because they just work incredibly well and they sound absolutely amazing!

Since originally owning the MR1000 and now 2 rack mount MR2000S's and using each very extensively - along with my PCM Apogees - I have found that utilizing DSD ADDS to and ENHANCES the quality of my audio recording experience, every single time. I believe they developed the existing technology with exceptional implementation in mind, in order to produce professional recording units that capture the highest quality audio possible - meaning the most accurate digital representation of the analog audio signal coming into the unit. And, fully knowing that to achieve this goal and implementation, they would NOT include the standard PCM decimation, etc. to multi-bit in the process - because there's a very real and negative sonic trade-off when doing so.

Of course most all companies need to market their products. But, non-engineered implies they're just shoveling previously produced metal & chips into a box and marketing "future proof" snake oil... trying to pour it down people's unsuspecting throats for a quick buck. That's what I interpret as sort of silly... please no offense meant.


==========================================

I have a 2-part question, if anyone would like to venture any answers/comments that would be cool. And, then I just have a final comment in this (already to long post of mine) FWIW. Again, my apologies if this detracted from this EXCELLENT thread in any way. I just think it all ties together and most all of us are simply seeking the best tools we can use (and afford) for the love of our craft and the soul-enriching creation and playing of MUSIC.

==========================================


Question: Taking DAW audio editing requirements out of the equation for a more apples-to-apples thought experiment, if you will (for those that haven't even heard these Korg DSD units)...

It's a live rock band playing in a great room, you've got a mixer setup and you're capturing a stereo track in both Korg DSD @ 5.6 MHz/1-bit and lets say a Lavry PCM @ 96k/24-bit AT THE SAME TIME. When playing back and comparing the 2 stereo digital files recorded in DSD (5.6 million sample a second) and PCM (96 thousand sample per second), which would you expect to sound closest to the SOURCE?

And, secondly, wouldn't you necessarily expect "some" audible, negative sonic alteration from the PCM converter that is performing all those extra processing steps (decimation, anti-aliasing, etc.) in order to convert the original hight sample rate digital capture into multi-bit PCM (that is required if you want to do any DAW editing)?


In any case, I would just encourage folks to at least trial and LISTEN to these Korg DSD recorders for themselves. Don't be necessarily put off by many of the negative comments. They truly do work as intended and the capture quality is STUNNING.

All the best!

P.S. I'm in no way, shape or form affiliated with Korg or any audio products company.

[doom64]

Originally Posted by string6theory

In any case, I would just encourage folks to at least trial and LISTEN to these Korg DSD recorders for themselves. Don't be necessarily put off by many of the negative comments. They truly do work as intended and the capture quality is STUNNING.

All the best!

P.S. I'm in no way, shape or form affiliated with Korg or any audio products company.

I'd be willing to try and listen to the recorders. Unfortunately I don't have the resources for such a luxury.

One of the things I always find funny with certain audio companies is lack of audio clips or audio clips within product videos. To me it's silly to not include those when selling an audio product.

You actually just gave me an idea that I may put together...Something that has been bothering me for awhile that needs to be addressed or at least attempted.

[Jon Hodgson]

Originally Posted by string6theory

My view is that Korg must have put quite a lot of engineering resources into these MR unit's and their implementation. Why? Because they just work incredibly well and they sound absolutely amazing!

There were no great engineering resources needed, no more than for any similar PCM recorder.

Korg didn't invent DSD, which in itself was hardly invented, once you have sigma delta converters (which were already the industry standard when DSD was introduced) bypassing the decimation filters and simply storing the output of the quantizer directly is a fairly obvious thing to try.

Korg didn't develop the converters in the MR units either.

It's an off the shelf ADC and DAC, with a DSD output and input respectively, to record it you simply write the bitstream to disk, job done.

So in terms of audio performance the only areas Korg had any leeway really are in the analogue front end, and in the software to convert to multi bit PCM (and the latter is just SRC, you lowpass filter and downsample, so no research to be done there, just a question of how much processing power you want to throw at the filters).

[Jazznfunk]

Yesterday i made simple test. Reaper with it's built-in tone generator via Voxengo SPAN, ASIO'ed thru old good/bad digi 002 line out in semi professional oscilloscope. Sorry for poor OSC pics, but they're completely understandable.

=======================================================================

TEST #1 - 20 000 Hz sine @ 44.1 kHz - screenshot and photo





This test proves, that 44.1 kHz is enough to sample 20 000 Hz sine. In picture sine looks pretty good. Nyquist is right.

=======================================================================

TEST #2 - 9600 Hz triangle @ 44.1 kHz - screenshot and photo





Triangle looks like non linear sine.
_________________________________________________________________

9600 Hz triangle @ 48 kHz - screenshot and photo





Triangle still looks like "sinish" in upper and a bit "trianglish" in lower waveform part.
_________________________________________________________________

96 Hz triangle @ 96 kHz - screenshot and photo





Still not perfect, but looks like a triangle.

Where's Nyquist?

[Jon Hodgson]

Originally Posted by Jazznfunk

Still not perfect, but looks like a triangle.

Where's Nyquist?

Well the first issue with this test is that the frequency graphs on Voxengo SPAN seem to be showing some pretty awful triangle waves, with no real consistency as to how bad they are.

The first is aliased like crazy, the second shows a second harmonic (which you don't get in a triangle wave, but since there seems to be a DC offset it looks to me like one side is clipping), the third at least shows a decent third harmonic, so I'm not surprised it looks the best.



a 9600kHz triangle wave should look like a sine wave on a 44 or 48 kHz system, on a 96k system there's room for one harmonic, so it should be rather more triangle like.

[Jazznfunk]

Thank You for the answer and explanation! Should i use other tone generator?

[DistortingJack]

People forget that we cannot actually "hear" triangle waves. The harmonics needed for a triangle wave to look like a triangle are all supersonic, just like we cannot hear a square wave either.

On a triangle wave of 9600 we can only hear the fundamental and the 1st harmonic (9600 and 19 200), which doesn't look like a triangle but sounds exactly like one. The 3rd harmonic is at 28 800 Hz, you can't expect to hear that!
Basically a properly filtered analog triangle wave will look exactly like the digitally sampled one on your picture.

By the way, the 44.1 kHz sampled one is really really messy because it's aliased. The algorithm you need to use needs to use internal filtering in order for the signal to be output correctly.

[UnderTow]

Originally Posted by Jazznfunk

Thank You for the answer and explanation! Should i use other tone generator?

I'm not sure what you are trying to test. As others and myself have clearly explained, harmonics outside the bandwidth of the digital audio (half the sample rate more or less) should be filtered out. These are not audible to humans so there is nothing missing to our ears!

I suggest you reread the first post in this thread.

Apologies in advance if I misunderstood what you are testing.

Alistair

[FredFraikin]

Originally Posted by MediaMix

We work at 24/44 or 24/48. But I never argue with someone who says they work at 88.2/96 because it sounds better. Whatever floats their boat.

Another question. So working in the daw at 88.2/96 is a waste of resources? Many have stated that plug-ins sound better at the higher sample rates and the overall sound is better, including many of the developers here like Cytomic. Is this a myth as well?

No it's not a myth! I totally agree with the fact that recording at 44.1/48khz makes no audible difference to recording at 88.2khz/96k but...but the fact is some plug ins do sound better at higher sample rate, there's a difference even from 44.1khz to 48k. It's clear on a heavy session, try by yourself same mix at 88.2k than convert it at 44.1k and listen to both print. The difference is there and obvious the more plug ins you use

[popmann]

I thought about this thread watching the last Newsroom on HBO...when Munn's character was talking about the debate of the debt ceiling being enough to cause the devaluation of the dollar...and in the same episode, someone explaining how he had to steer online discourse in an absurd direction to get into some kind of troll club.

Feel free to continue this riveting discussion based in a reality alternate to mine.

[string6theory]

Originally Posted by Jon Hodgson

There were no great engineering resources needed, no more than for any similar PCM recorder.

Korg didn't invent DSD, which in itself was hardly invented, once you have sigma delta converters (which were already the industry standard when DSD was introduced) bypassing the decimation filters and simply storing the output of the quantizer directly is a fairly obvious thing to try.

Korg didn't develop the converters in the MR units either.

It's an off the shelf ADC and DAC, with a DSD output and input respectively, to record it you simply write the bitstream to disk, job done.

So in terms of audio performance the only areas Korg had any leeway really are in the analogue front end, and in the software to convert to multi bit PCM (and the latter is just SRC, you lowpass filter and downsample, so no research to be done there, just a question of how much processing power you want to throw at the filters).

Granted Jon , I would think (& hope) most folks already know that converters from most all companies, like Apogee, Lavry, Korg, etc. are designed and engineered from mostly pre-existing tech and componentry, and then fabricated into their end product. The same may be said for computer companies, like Apple or Dell, who design and assemble pre-existing chips and componentry from Intel, etc.

There will always be much vertical and horizontal, inter and intra-industry synergy, sourcing & integration. Company A specializes in producing the amazing "X-chip" and companies B, C & D all use this X-chip in their engineered designs, but may develop differing implementations within those designs (using more or less engineering resources). And, of course, industry politics, patents & profitability, among many other factors, play crucial roles (especially wrt standardization).

But, if someone suggested Lavry converters or Apple desktops are just marketing products and not engineered products, I'd think that was silly as well. One could argue the relative levels of proprietary (or otherwise) implementations, but that certainly wasn't my point.

I've just read so many forum comments from "inter-industry" folks regarding the Korg MR units, wrt the quality/usability/results/intentions of their proprietary 5.6 MHz DSD audio implementation, framed (crudely to eloquently) in such a way as to be ultimately or entirely "dismissive". A quick GS site search on DSD would make this clearer. Many of the same respected folks would keep popping in DSD-related threads and "dismissing"... to intentionally(?) or otherwise dissuade people.

Anyway, I'm glad that I personally didn't give those many dismissive comments too much weight (as I interpreted a common thread of unwarranted bias in them). That's what I was just passing along, as well as my personal and VERY POSITIVE experiences using the Korg DSD units, FWIW.

BTW, it's good that you brought up the Korg software (AudioGate). It's definitely an indispensable part of the MR units' product package and does indeed add a lot of value. I thinks works incredibly well for SRC-ing DSD to PCM.

Warm Regards,

[antstudio]

A lot here about recordings at 96 sounding better (or not) than 44.1. But how do you actually do a valid test to compare? Listening to my lowly 320kbps MP3s when my Apogee Duet 1's output is set to 96khz, the sound is much fuller, smoother and the bottom end is much more distinct and clear than when set at 44.1 or 48. Obviously the recordings are exactly the same (and at 44.1 not at 96). Internally I'm assuming there's a conversion from 44.1 to 96 before the D/A. I just think the 96khz D/A in the Duet is better then the 44.1/48khz D/A. It's so obvious, when once in a while a random app changes the Mac's audio rate to 44.1 without my realizing it, as soon as I play anything I immediately realize and open Audio Midi Setup to switch it back.

..ant

[hugol]

Originally Posted by FredFraikin

No it's not a myth! I totally agree with the fact that recording at 44.1/48khz makes no audible difference to recording at 88.2khz/96k but...but the fact is some plug ins do sound better at higher sample rate, there's a difference even from 44.1khz to 48k. It's clear on a heavy session, try by yourself same mix at 88.2k than convert it at 44.1k and listen to both print. The difference is there and obvious the more plug ins you use

Yep - although obviously some plug-ins oversample internally or employ other techniques to minimise the aliasing we're talking about here. So I think it really depends what plug-ins you're using as to how significant it is.

I guess the distinction here is we're not recording / playing-back audio untouched - we're performing significant processing on it.

[FredFraikin]

Originally Posted by hugol

Yep - although obviously some plug-ins oversample internally or employ other techniques to minimise the aliasing we're talking about here. So I think it really depends what plug-ins you're using as to how significant it is.

I guess the distinction here is we're not recording / playing-back audio untouched - we're performing significant processing on it.

Absolutely!

[doom64]

Originally Posted by FredFraikin

No it's not a myth! I totally agree with the fact that recording at 44.1/48khz makes no audible difference to recording at 88.2khz/96k but...but the fact is some plug ins do sound better at higher sample rate, there's a difference even from 44.1khz to 48k. It's clear on a heavy session, try by yourself same mix at 88.2k than convert it at 44.1k and listen to both print. The difference is there and obvious the more plug ins you use

+1. Digital signal processing likes more samples to work with. Ask plugin developers and they'll tell you. And up/oversampling is no substitute for a high samplerate original.

I got bored today and resampled some of my 96 kHz recorded files. Replaced old mix files with the resampled files and rendered two different mixdown files. The control was a 64-bit floating point bit depth (my DAW's native mix format).

Then I phase-aligned the tracks in a new project and cut the regions up in 10 second or so chunks. I closed my eyes, clicked my cursor randomly and hit play. Whenever I heard the sound change I marked it with my keyboard. When the song was done I opened my eyes and I was right on each change.

The 96 kHz mixdown sounded more "analog" for lack of a better word. It had better separation, it sounded rounder. In contrast the 44.1 kHz mix down had "grunge"for lack of a better work. It had a harsh/edgy quality to it that the 96 kHz mixdown didn't. The soundfield also sounded flatter/not as wide. Separation wasn't as good and I believe the volume level changed slightly as well.

This was on a simple vocal/acoustic guitar arrangement so I can only imagine the difference it would make on a mix with more than 2 tracks.

The test wasn't 100% fair because the files were sample rate converted but I used an extremely high quality SRC program for the test.

As an aside, I did two other mixdowns and null tested them. 64-bit/96 kHz and 64-bit/44.1 kHz, mixed with the 96 kHz originals. With my DAW's built-in SRC there was more of a difference but with my third-party SRC program it was much less. About a -35 average level null with the differences being in the 14 kHz and above range. Inperceptible on my prosumer Klpsch computer speakers, which are the same ones I used for the other tests.

[Jon Hodgson]

Originally Posted by doom64

+1. Digital signal processing likes more samples to work with. Ask plugin developers and they'll tell you. And up/oversampling is no substitute for a high samplerate original.

I got bored today and resampled some of my 96 kHz recorded files. Replaced old mix files with the resampled files and rendered two different mixdown files. The control was a 64-bit floating point bit depth (my DAW's native mix format).

Then I phase-aligned the tracks in a new project and cut the regions up in 10 second or so chunks. I closed my eyes, clicked my cursor randomly and hit play. Whenever I heard the sound change I marked it with my keyboard. When the song was done I opened my eyes and I was right on each change.

The 96 kHz mixdown sounded more "analog" for lack of a better word. It had better separation, it sounded rounder. In contrast the 44.1 kHz mix down had "grunge"for lack of a better work. It had a harsh/edgy quality to it that the 96 kHz mixdown didn't. The soundfield also sounded flatter/not as wide. Separation wasn't as good and I believe the volume level changed slightly as well.

This was on a simple vocal/acoustic guitar arrangement so I can only imagine the difference it would make on a mix with more than 2 tracks.

The test wasn't 100% fair because the files were sample rate converted but I used an extremely high quality SRC program for the test.

As an aside, I did two other mixdowns and null tested them. 64-bit/96 kHz and 64-bit/44.1 kHz, mixed with the 96 kHz originals. With my DAW's built-in SRC there was more of a difference but with my third-party SRC program it was much less. About a -35 average level null with the differences being in the 14 kHz and above range. Inperceptible on my prosumer Klpsch computer speakers, which are the same ones I used for the other tests.

Even something as simple as an EQ can act differently on a different sample rate. Due to the way the fiter response curves from the equivalent analogue response the closer you get to Nyquist. This doesn't necessarily make one better than the other (it's not distorting, it's just a different shape of EQ curve), but it does mean that the same settings could well sound different.