i am interested if anyone has any further details of the study done at the 2010 aes seminar into samplerate descrimintation.

i read constantly online by people that 44.1khz is as good as need be used because of nyquist, human hearing range etc.

so i was interested when i discovered an aes paper where abx blind tests showed a difference between 88.2 khz and 44.1 kHz recordings.

How Do We Evaluate High Resolution Formats for Digital Audio?
Chair: Hans van Maanen, Temporal Coherence - The Netherlands
Panelists: Peter Craven, ,Milind Kunchur, University of South Carolina - SC, USA
Thomas Sporer, Fraunhofer Institue for Digital Media Technology IDMT - Ilmenau, Germany,
Menno van der Veen, Ir. Bureau Vanderveen, Wieslaw Woszczyk, McGill University - Montreal,
Quebec, Canada

Abstract: Since the introduction of the High Resolution Formats for Digital Audio (e.g. SACD,
192 kHz / 24 bit), there has been discussion about the audibility of these formats, compared to the CD format (44.1 kHz / 16 bit). What difference does high sample rate and bit depth make in our
perception? Can we hear tones above 20 kHz? Can we perceive quantization errors in 16-bit audio? Does high sample rate make a difference in our phase resolution? Are we even asking the right questions? Controlled, scientific listening tests have mostly given ambiguous or inconclusive results, yet a large number of consumers, using "high-end" audio equipment, prefer
the sound from the "high resolution" formats over the CD. The workshop will start with introductory notes from the panel members, discussing the differences between "analog" and first-generation digital formats, address some of the paradoxes of the CD format, present results on "circumstantial" evidence and subjective testing, show results on the audibility of the human
hearing, which cannot be explained by the commonly accepted 20 kHz upper limit and discuss the problems and pitfalls of "scientific" listening tests, where possible illustrated with demonstrations.

These introductory notes should provoke a discussion with the audience about the audibility of the improvements of the "high resolution" formats We attempt to reach consensus, where possible, regarding what is known and what is not with respect to our ability to perceive the differences between standard and high resolution audio. We further discuss the paradigms of
testing for evaluating the quality and perception of high resolution audio, how to structure the tests, how to configure the testing environment, and how to analyze the results.
The outcome of the workshop should also be to find the way forward by identifying the bottlenecks which—at this moment—hamper the further implementation of the "high resolution" formats for "high-end" audio as these formats create an opportunity for the audio industry as a whole: better sources stimulate the development of better reproduction systems.
These thought provoking presentations gave some teaching for psychoacoustic test methods and some fascinating recent results on perception thresholds. Peter Craven gave an insight into subjective testing and how the forced decision ABX test may in fact fail to find out what the ear /brain perception is doing, where the test blocks the natural perceived response to audio quality variation unless the differences are relatively gross.
Milind Kunchur outlined the extreme care necessary to establish sensitive tests to establish a 5uS or so temporal detection threshold, backed by a theoretical analysis of this aspect of hearing.

5.Finally in the paper sessions, ‘AES London 2010 P18 - Audio Coding and
Compression’ there was a presentation on the audibility of differences between 88.2 and 44.1kHz recordings with statistical verification of the preference for the higher resolution, whether a recorded original or when compared with a carefully down sampled version of the Hi Res material.

The abstract and reference is given below:
P18-6 Sampling Rate Discrimination: 44.1 kHz vs. 88.2 kHz—Amandine Pras, Catherine Guastavino, McGill University - Montreal, Quebec, Canada
It is currently common practice for sound engineers to record digital music using high-resolution formats, and then down sample the files to 44.1 kHz for commercial release. This study aims at investigating whether listeners can perceive differences between musical files recorded at 44.1 kHz and 88.2 kHz with the same analog chain and type of AD-converter. Sixteen expert Listeners were asked to compare 3 versions (44.1 kHz, 88.2 kHz, and the 88.2 kHz version down-sampled to 44.1 kHz) of 5 musical excerpts in a blind ABX task. Overall, participants were able to discriminate between files recorded at 88.2 kHz and their 44.1 kHz down-sampled version.
Furthermore, for the orchestral excerpt, they were able to discriminate between files recorded at 88.2 kHz and files recorded at 44.1 kHz.Convention Paper 8101
Comment:
Taken as a whole these reported developments suggest that Hi Res audio is beginning to take off, its credibility is clearly improving, and we have the means to deliver it to audiophiles via a low cost and now widely available carrier. About 20 titles have been announced so far with intent for 20 more from Naxos and Sony. Web delivery of Hi-Res surround sound is clearly impractical, the audio Blu-ray disc has the potential to deliver value to all concerned with genuinely higher quality sound, stereo and surround.

surprised no-one has any further info on this????

Sounds like the down-sampler needs work. If they want to do something useful, they should go back and shoot out down-samplers, including DAC to ADC, because 99% of the public is going to hear the 44.1 only. And they've proven 44.1 is perfectly capable by their own test (4 out of 5 pieces indistinguishable).

Because engineers are encouraged/forced to use 96k rather than 88.2, 48 or 44.1 because of the existence of Blu-ray, they should provide more information on just what they used to "carefully down-sample", and go back and do it again with more than one down-sample option.

i have no doubt some of what you say is true as to why samples rates get used. it could also be said that some engineers use higher sample rates because they prefer the sound of it and some engineers use lower sample rates because of workflow constraints.

have you read the paper?

I always try to go as large a file as possible before bouncing down. I find it affects headroom either way, so even if it's subtle it's for the best. I would love to work with larger track counts at 88.2Khz but MADI isn't there yet.

No I didn't read the paper. I made all of my conclusions from the piece you posted. But I found that they discussed this at hydrogenaudio three years ago. In that piece they added specifics from the paper, including the fact that they used Pyramix to downsample.
Google " hydrogenaudio sampling rate discrimination " to find the thread.
The fact that they could differentiate only the orchestral piece, who knows what to make of that.

That strikes me as the correct interpretation of the results.

Recorded at 88.2k and downsampled to 44.1k = mostly distinguishable.

Recorded at 88.2k and recorded at 44.1k = mostly indistinguishable.

Those outcomes imply that the effect of the downsampler is what's clearly audible.


Gregory Scott - ubk

i can't believe that anyone is still debating this, saying one cannot hear differences between 44.1k and 96k is like bragging about how poor monitoring a person has, or how bad acoustics one is sporting, it is as easy as picking what compressor to use, or what pants to wear.

believe what you want, but there is a reason why the chip makers have 192k as an option.

just on the downsampler.

isn't the effect of the downsampler or even the output circuit always in the way.

also, what about sampling density, won't that also have an effect.

Please re-read what Gregory wrote, because it pretty much contradicts the OP's assumptions.

Recordings at unitary and 'double' rates are mostly indistinguishable from each other in the results of the referenced study. However, the result from sample rate conversion from 88.2 to 44.1 WERE mostly distinguishable -- indicating that it was the sample rate conversion that was causing most respondents to recognize a difference -- NOT a qualitative difference between the tested units tracking at 44.1 v. 88.2 -- understood?

So the one conclusion that is supported by these test results is that those using similar audio converters and sample rate conversion would likely do better simply recording at 44.1 kHz to begin with.

THIS is a very simple logic and if anyone doesn't get it, they need to send themselves to the audio re-education gulag for re-entrenching.

It's safe to say it's pretty much a sales gimmick. You should check out what the lead design engineer at Prism had to say about DSD and 192 kHz. Should be a real eye-opener for many. (I can't find the article right now; it was in an audiophile mag someone linked to here [trying to prove conclusions that the interview didn't actually support] and he's a bit discreet about saying what he says -- that such formats don't really offer any sonic improvements and that quad rates are in the vicinity of the threshold of inaccurate ADC processing windows and that Prism really only offers those features in some top end units because they feel such a unit should be able to deal with virtually any format a serious music collector might come across.)

Very interesting.

The authors of the conference presentation you cite appear to have given up on that research., Guasativno's website, which lists her cv in detail, shows no work on sample rates and perception of quality. You cna by a copy of the conference paper if you like:

http://www.aes.org/e-lib/browse.cfm?elib=15398



Along with Pras, her co-author she did do a very interesting interview with six working commercial producers who they did not name:

"Five producers (all except CI3) commented on the technical advances in digital technologies in the last two decades. On several occasions since the early 1980s, PI6 had the opportunity to compare the quality of analog and digital equipment and in the past few years he observed almost equivalent sound quality between the best digital and the best analog equipment. CS1, CS4, and CS5 focused on the amazing improvements in digital tools for editing and correcting, such as the possibility to change the tempo of a take. Moreover, PI6 mentioned the possibility of doing revisions at any point in time, which was not possible with analog equipment. CS1 and CS5 explained that musi- cians became aware of these possibilities and that they now have higher expectations in terms of postproduction. However, all interviewees noted that producers do not use the full capabilities of these tools due to time constraints: the irony is that we have all these tools so we can spend all this time to make these recordings better and better, and it is great, but nobody can pay for all this time (CS5).
"

we are talking about digital.

firstly i[m not arguing with Gregory. so there is no contradiction here. you will notice that i said
"isn't the effect of downsampling always there".

also,

maybe you've read a report i haven't or in fact the rest of the report. because the only information i have is the report i posted above and it says.

"Furthermore, for the orchestral excerpt, they were able to discriminate between files recorded at 88.2 kHz and files recorded at 44.1 kHz."

it doesn't say recorded at 88 and recorded at 41 is indistinguishable.

all i ask is actually read stuff. stay calm and try not to put words in peoples mouths.

Why would you think that downsampling is 'always there'?

If you track at the sample rate you're releasing at, there's no need to downsample.

What were you thinking?

Yes, and I'm doing you the favor of looking for the paper you mentioned. I gave you a link, you can buy it and read it. The authors of the paper I quoted were also looking at digital

thanks, have seen the aes link and did not want to pay for it which is why i started this thread in the first place in case someone had the paper so i could read it.

i am nervous of you and others, turning this into a digital analogue argument because of what has happened in previous threads so i pulled you up on it by pointing out that posting stuff on analogue editing methods had nothing to do with the thread.

thnaks

You quoted a spambot. There was no "poster" - it was an excerpt from an earlier post (herecomesyourman's to be precise). The resultant bickering was effectively sparked by a machine!

As to the original premise; I think again you're kind of taking a generalisation and arguing against it. Personally, if I had the resources, I'd always work at 96k. However, I don't feel processing power is up to it yet. The slight improvement you might get from working at a high sample rate is weighed down by the workflow compromises for me.

with a "perfect" converter, 44.1k might well be indistinguishable from 96k. We don't have a perfect converter, so a higher sample rate makes perfect steep filters less necessary, and thus I guess sound accuracy might be better.

Sample rate has nothing to do with headroom. That's bit depth.

Meh. I just mean a larger file. The larger the sample rate on the master before you bounce down the better...even when you get down to MP3 the file is on average larger than it would be from starting a lower sample rate first.

The more information, the closer things generally sound to what I'm playing back in the DAW. This is why moving to a DSD recorder has been even better for me than internally bouncing via real time bounces, or recording through buses, or through busing back through my ADA onto a new track when I record.

(Side note: Wow...what's with all the deleted posts?)

As it says - a bunch of off topic bickering, started when someone replied to a spambot

ah, it's a pity you took down the charming commentary.

i was just about to answer the final question asked.

LOL!

A schismogenic spambot.

Sort of a cyberzombie agent provocateur.


It just goes to show... something or other. G'night all. It's rounding up on midnight in Cali.

Ok.

I'm intrigued at the allegation that downsampling from 88.2 to 44.1 is causing distinguishable damage. In theory, this is the easiest SRC to perform because it is exactly double. Surely all that is necessary is to discard every second sample, so it's a bit hard to screw up the maths. But if this allegation is true, it must be something to do with the filtering. 88.2 will capture energy above 22 kHz which will have to be filtered out when down-sampled, whereas if it was recorded at 44.1 then it would not exist right from the start.

what the shits a spambot?

Spambot - Wikipedia, the free encyclopedia

"Other spam messages are not meant to be read by humans, but are instead posted to increase the number of hyperlinks to a particular web site, to boost its search engine ranking."

That's what those broken link things are...if you see one, report it and we'll remove it!

samplerate doesnt affect headroom. and its not what you believe or find, its what is fact that matters.

End of the day if I started with a larger file before compressing down to Wav or MP3...for whatever reason things sound better / seem to feel more alive even at high volumes, etc.

MP3's are what? 11% of the original file at 128kps? Seems to me it's better to start as big as possible. (As long as your computer can take it.)

that's been my experience as well. the higher the sample rate the more I seem to like what I hear.

I'd love to hear quad dsd at this point.