Some generally great sounding sample rate converters (SRC) alias >22kHz content into the <22kHz range at high energy levels. Although my hearing caps between 17-18kHz, I asked myself wether this could influence the playback of lower frequency content on real world systems.

If you want to try this yourself, convert the following sample with your favorite SRC to 44.1 kHz (-70db sine wave sweep vs. -20db modulated sine wave at 23333 Hz +/- 1000 Hz):


96 kHz, 32 bit float: Click to view attachment

Ideally it should look like this afterwards:


44.1 kHz, 32 bit float: Click to view attachment

Anyway, try to ABX your conversion of Sweep_70db_HF_20db.wav vs. Sweep_70db_441.wav. I got anything from outright terrible (VMware Fusion's internal SRC sounded like amateur radio) to indistinguishable (Sox VHQ).

The following pair doesn't sweep and this should make it easier for some ears:

96 kHz, 32 bit float: Click to view attachment 44.1 kHz, 32 bit float: Click to view attachment

Remember, these are synthetic test samples specifically made to be tough on sample rate conversion. If a SRC fails this, this doesn't mean that it couldn't still produce excellent results for normal music!

If you are generally interested, wether audible effects of very high frequencies above your threshold of hearing are possible, try to ABX the following two very extreme samples against each other (-80db 4000 Hz sine wave vs. -6db 21000 Hz +/- 500 Hz modulated sine wave):


Redbook compatible, triangular dither: Click to view attachment


Redbook compatible, triangular dither: Click to view attachment

This post referencing this site had initially drawn my interest in the matter. This thread's samples should put a much harder burden on the tested SRCs.

Since real world amplifiers and speakers aren't totally linear, some volume differences could apply. Maybe someone with more insight into the physical side of this could explain wether that matters.

What does a bad job of resampling look like?

(I'm, a long term user of CEP 2.1 and so I have very little experience with sample rate conversions done wrong.)

You could go to the infinitewave site and have a look at the "Ableton Live 7" results.
An excellent pulse response but lots of aliasing.

that's a horrible impulse response! Should like like a sinc wave (sin(x*pi)/(x*pi))

I disagree - it is a wonderful impulse response. It looks like an analogue impulse. Which is exactly what any converter will try to achieve.

A bad downsample looks like this one done with the internal resampler in WaveLab 5:


A good resampler (iZotope RX Advanced) looks like this:


... when it is configured like this:


However, I subjectively prefer this setting (with music):


... which results in this:


Not perfect, but still no aliasing artifacts mirrored below 20 kHz - and wonderful impulses.

I'm afraid that's not possible. At least, linear interpolation is not suitable for high quality conversion. So I'd rather prefer impulses like this:

For everyone still using Cool Edit Pro 2.1 or Audition 1.x: this is very similar to what you get with the following settings: Quality: 999, Pre/Post Filter: OFF, Resolution: 32 bit.

Turning the pre/post filter on just lengthens the impulse response by a factor of 4 or so, so I always leave it off.

Chris

This reminds me of my grandparents' old wallpaper. It seems that they used some ancient form of resampling to design it..

No, Audition is pretty different from what Cavaille has plotted. Quality=999 is much steeper, and Audition has -6 dB point at Nyquist while Cavaille's is -36 dB.

I´d prefer the left graph. When I configure the SRC to resemble the graph on the right side it sounds remarkably different and "strange". But then... this could be very useful for certain things.

I don't think it does. You must be doing something wrong or it's the placebo effect in action. Please look at the time scale. The post ringing is under 2 milli seconds and the range of group delays within the audible band is probably limited to a couple of micro seconds. I would not expect you to be able to hear a difference.

Cheers!
SG

Should I do an ABX and post the samples I used for it? That should prove it... I already told that in another thread that I found out for myself what the "best" upsampling method is in order to reach the sound of the original 24/96 source. I used my ears and several ABX for it and I was happy when I finally wasn´t able anymore to distinguish the upsampled file from the original 24/96 file. I offered some upsampled files here. I also described that I came up with this by testing the SRC from iZotope. In order to find out what sounded closest I also experimented with impulse setting or Pre-Ringing. And when I had no pre-ringing but huge post-ringing it sounded "strange", as if someone had been playing around with some stereo expanding tool. The same with downsampled material. I didn´t try testtones like this one here in this thread, I used only music.

Yes, Alexey, sorry, I realized that while you were typing your response and pulled back from "virtually identical" to "very similar"... which I guess is still an overstatement, I remembered my Audition experiments wrong. Quality = 600 or so is closer to what Cavaille came up with. But I prefer the whole 999

Chris

99.9% of the golden ears who have been educated by our favorite golden-eared journalist(s) would think they would prefer the Abletron impulse. ;-)

The interesting question is what this sort of implulse response does to music. If I downsample regular music to say 11 KHz sample rate, would I prefer the Abletron or CEP downsampler?

Good question. Does anyone have Ableton and can upload a 11.025 kHz downsampled (using the High Quality setting) music item? Like castanets? Or the Everything is Green (eig) sample?

Chris

But why high quality? These pictures at post #3 are from "Ableton Live 7", not "Ableton Live 7 High Quality".

Well, the non High Quality is without any anti-aliasing filter. I think it's not hard to imagine that this will produce pretty bad sounding 11-kHz audio, especially for tonal signals. IIRC, there is such a demo on the CD in this book. For me, it would be more interesting to hear what their "quality" mode sounds like, as this still produces significant aliasing.

Chris

Arnold, you know that I read some of those magazines from the "golden-eared journalists". And in theory, this impulse resembles an analogue impulse pretty much. But what most journalists fail to see is that these impulses only affect frequencies close to nyquist. To my knowledge, the effect on the rest of the frequency spectrum diminishes to zero the lower you get. Still, since being exposed so many years to the "impulse-lie" I tend to respect it - and only to make sure. Furthermore, since I came up with my up- and downsampling technique by ABXing the results, I´ve got these better impulses only by chance. Now there are two possibilities: 1. the SRC from iZotope colours the sound 2. despite theory, impulses can be heard.

In the case with Ableton the theoretical advantage of having a perfect impulse is completely destroyed and overshadowed by strong aliasing artifacts mirrored back into the audible signal. So there appears no benefit in using this SRC.

Such "perfect impulse" will inevitably lead to very poor artifacts. Linear interpolation is poor for both audio and video.

Hopefully you understand that those two facts aren't unrelated either. The impulse response and the frequency response are just two sides of the same coin, they have an underlying mathematical relationship that can't be avoided. In fact you can prove that if the frequency response is finite then the impulse response must be infinite (as in infinite extent) and visa versa. Note here that we are talking about "finite extent" in an exact mathematical sense as in going precisely to zero outside of some finite domain. Of course an impulse response can have an "infinite" extent in a strict mathematical but still sensibly go to "zero" in an audible sense.

That´s why it is so difficult to program decent SRCs I believe. It´s perfectly simple: If you have a perfect impulse you´ll sacrifice other things by introducing audible aliasing artifacts. And if you don´t have any aliasing, you basically have to choose between pre- and post ringing or no pre-ringing but huge amounts of post-ringing. It´s the same with every SRC around... That´s why I like SRCs being able to be "tweaked" - that way you´ll get the best of "both worlds" by adapting the SRC to the needs of the job.

You're really ascribing a lot more importance over temporal masking than I believe is generally recognized. Minimum-phase conversion is a rather high-end topic. To the best of my knowledge, nobody (yet) has ever ABX'd a difference between two correctly implemented (ie, not Ableton Live) converters. So you do have to excuse us for being skeptical.

If you're going to keep asserting a positive 24/96 test, one of these days you will need to actually post the details on it. We won't be that hard on you, will we?

Also: It's worth pointing out that the "perfectness" of the Ableton filter is really an illusion caused by incorrect rendering of the waveform.

RX plotted the waveform quite correctly, it uses a high-order windowed sinc interpolation to reconstruct the analog waveform. The reason why there's no ringing is that the sampling rate of the impulse response is 14.112 MHz (this is the rate at which SRC filter works). If this is plotted at 44.1 kHz, the ringing will be definitely present.

Oh, that is interesting. How can I setup the waveform window in RX to show me the impulse response and what tone do I need for that? I always wondered how people were doing it...

It's not even a perfect impulse in the time domain - it might look like one for the specific 100% artificial example where you're looking at an impulse perfectly co-timed with a sampling instant - but delay the signal by half a sample, and then see what you get

Cheers,
David.

David, sorry, you are on the same track of misunderstanding with Axon. Let me explain once again. The pulse plots at InfiniteWave show the amount of ringing introduced into the signal at all possible sub-sample shifts, not at the only shift when the impulse is perfectly aligned with a sampling instant. The depicted filter includes all the polyphase components of a multirate SRC filter. It is in fact a multirate SRC filter operating at 14 MHz. See below.
Indeed, Ableton is using a linear interpolation, hence no ringing is supposed to be introduced in the output signal, no matter what the "phase" is.
Ringing of SRC should not be confused with ringing of a possible further D/A conversion. This second ringing is completely defined by an oversampling filter of D/A and has nothing to do with the tested SRC. The graphs at InfiniteWave are only concerned with ringing of SRC.



That's all about preparing the waveform data, not about setting RX. The pulse train test signal (see InfiniteWave Help and FAQ sections) is used to extract all the polyphase components of the multirate SRC filter operating at 14 MHz. Then this filter is written into a 14-MHz WAV file, and this WAV file is opened by RX. We do not even need to change default RX display settings.
However, for those interested, RX has a selectable interpolation order for showing the "analog" waveform between digital samples (see Preferences/Misc). The default value is 30 (windowed sinc), but it can be set to 0 or 1 for more accurate display of impulse responses of digital filters.

I didn't realise the testing was so sophisticated, and I've just enjoyed reading a little about it.


However, that wasn't my point at all - while you have gone to great lengths to accurately show the impulse response of these SRCs, a normal reader might conclude that what is shown is the response of a given SRC when fed with an(y) impulse.

In reality, there isn't just one impulse "response", but a range of responses, depending on the time relationship between the input impulse and the output samples.

For a good SRC the differences are real but (arguably) irrelevant - after a decent DAC, the analogue signal is the same for any given original impulse.

For some bad SRCs, the differences are equally real but represent the destruction of the original signal. The result is that, after a decent DAC, the amplitude of the analogue signal can vary dramatically depending on the time location of the original impulse.


So it's not a case of "great impulse, poor sine wave" - both are mangled.

Cheers,
David.

You are right: there is a range of possible output impulses, when SRCing a single impulse. And all these possible impulses are contained in the "combined" impulse that InfiniteWave is depicting.

...which is fine for a time invariant system, but in the case of some "broken" (time variant) SRCs means you'll be hiding a fundamental problem?

(Not sure any of these SRCs are that bad, but it's possible).

(Not sure what I'd plot in this case - maybe some overlays?)

Cheers,
David.

Hi,



Not wishing to hijack the thread, just a quick question:
Is the difference between different resamplers (with regard to impulse response) similar to the kind of difference between different EQs (like minimum phase vs linear phase ?)

I already ABXd myself multiple times comparing minimum phase and linear phase EQing. I can reliably identify differences.
In bass frequencies I can percieve a phase shift a lot more than a pre-echo; it also impacts stereo imaging. Did not really compare high frequency EQing, perhaps I would tend to prefer less pre-echo there.

In this thread: minimum phase vs linear phase eq
I asked what are the thresholds of audibility of pre-echo and of phase shift. (Sadly no answers) But, perhaps this would also be useful for resampling ?
Has any comprehensive research been done on this subject ?

And finally, can different material benefit from different resamplers ?
If yes, would it be possible, and would it make sense to continuously, automatically adapt the resampling filter's impulse response to better suit the specific frequency spectrum being resampled, as to minimize audible pre-echo and phase shift ?

Currently, its seems like
1. Don't care, just resample
2. Guess/follow advice/theorize on which resampler is best
3. ABX different resamplers for one or a few songs, pick best, generalize and use for everything
4. ABX resamplers for every song (if differing material can benefit from different resamplers) which I am not crazy enough to do.

Surely there is a way to automate the process of finding the optimal resampling impulse response according to the specific material, but I doubt it has been done already. Would this make any sense ?

Thanks !

http://www.hydrogenaudio.org/forums/index....st&p=608302


And phase settings of a resampler are essentially phase settings of its low-pass filter. So you should really compare (very) high frequency EQing.

Similar. A resampler is supposed to make no changes to sound quality other than the obvious one implied by the downward change in Nyqusit frequency.



The difference being that eq should *generally* take place at an audible frequency, while due to the crazy world we live in, resamplers often make changes above the range we can hear.



Unlike high frequencies the huma ear actually has a mechanism for discerning changes in phase. If you don't apply a phase change equally to all relevant channels, there will be FR changes which can easily be audible.



If I remember that thread correctly, it was limited to resampling to/from Nyquists that were above audibility.



It's the same basic problem. Both A-D and resampling benefit from brick wall filtering as a general rule.



I don't know about exactly this problem, but certainly any of a number of relevant problems involving the same basic procedure - digital low-pass filtering.



Depends on what you call a benefit. Some people like the artifacts of a grab-bag of distortion, look at vinyl!

If you define the goal of resampling as I did above, then a reampler that is good for very many things even tough jobs, it should also be good for very many more things that it hasn't been tested with.



Only if you see resampling as a EFX.




The usual sitaution is that ringing is less audible when the pre/post ratio does the best possible job of hiding it below the relevant temporal masking curve, which says that the ear is more sensitive to pre-ringing than post-masking by something like 3-4:1.

IOW more post masking than pre masking would seem to be a reasonable goal.

The whole game of downsampling changes meaning when the Nyquist frequency is smack dab in the middle of the audio band, such as happens when we process for the best possible speech quality with minimal data.

How about these:

H. Møller and P Minnaar, "On the audibility of all-pass phase in electroacoustical transfer functions," J. Audio Eng. Soc., vol. 55, pp. 115-134 (2007 March).

J. Blauert and P. Laws, "Group Delay Distortions in Electroacoustic Systems," J. Acoust. Soc. Am., vol. 63, pp. 1478-1483 (1978 May).

Alexey,

I am intrigued by the nature of the iZotope intermediate phase filter.

When downsampling to 44.1kHz it is linear phase up to 14kHz or so,
and only then turns MP-ish.

Is there anything you can/want to say about its architecture?

Thanks,

Werner

The design goal was to be able to smoothly transition between linear-phase and minimum-phase filters. It not only allows trading pre-ringing for post-ringing, but also linearizes the passband phase response as much as possible. Our filter is a hybrid: it uses a linear-phase response below a certain frequency and a minimum-phase response above that frequency. The filter is FIR, so the design can be a straightforward window method.

So, does it sound any different than a more naively-designed filter?

For downsampling to 44 KHz sampling?

For downsampling to 10 KHz sampling?

Frankly, I have no idea. But our testers say that yes.

BWTW Alexy, you're the same Alexy Lukin whose name comes up when I click about in RMAA? If so, got any tips about what to do about the dozens of times I totally crashed the current downloadable version last night? Audio interface was a Card Deluxe with their current latest driver, and the OS was XP SP2. Older Via chipset with 333 Hz RAM.

May I kindly suggest that this is a rather exotic platform.

Arnie, I've been developing RMAA since earliest versions, but more recent versions have been done without my participation. I'm sure that developers of the current version (also mentioned in the About box) will appreciate your feedback.