Maybe, but they won't play low frequencies instead of high ones, like all AC97 soundcards seem to do.

I don't know what I've learned more about in this thread: human hearing, or average sound cards.

It is bizarre (and sad) that we have a general standard on PCs that means that most of the audio from PCs is wrecked in this way.

This thread wasn't intended to be an advert for non-resampling sound cards, but that's how it's worked out!

Thanks to everyone who voted and posted (you still can if you haven't already!).

This raises another issue for tuning audio codecs: if you remove very high frequencies when encoding, the owner of a good sound card may not detect this (because it's beyond the range of their hearing), whereas the owner of a bad sound card may detect it, because the frequencies were being aliased down into the audible range.

I love this board - so much to learn!

Cheers,
David.
http://www.David.Robinson.org/

what frequency would be at 4.5 seconds?

With all due respect, this poll is about the quality a National Inquirer poll has

You cannot expect to get reliable audiometric test results with unknonwn source, amplification and uncalibrated headphones in a normal noisy environment.

And yes, this may just as well include those of you who use top of the line Sennheiser headphones and 24bit/96kHz soundcards to play back the test tones.

For example, when you think you hear a tone of frequency XX kHz you might just as well be hearing a tone at ( XX - YY ) kHz for all we know. This could be due to intermodulation components caused by your test equipment (source, amp & transducers or a combination of these).

Further, even if you think you can't hear a tone at XX kHz it might be due to A) daily variance, B) improper test setup, C) bad test gear.

So, do not draw any too far fetched conclusions about what humans can hear or what you personally can hear based on this poll or your own bedroom testing (unless you are an audiometric testing trained and you know what you are doing and have access to proper test gear for ultrasonic auditory measurements).

Some of you might also be interested in knowing that there is no widely accepted international standard for hearing tests at beyond 15 kHz. Some experienced practitioners of audiometric testing go as far as to claim that most data about human hearing beyond 11 kHz is suspect and cannot be trusted.

The differences in level of audibility for single tones can cary at least +/-10 dB (Pa/m2 at the tympanic membrane) from day to day with the same test subject. This for high frequencies. That is, the intensity of sound can vary 20 dB and you'll still get the same test results about the audibility of test tones.

So, while this poll and testing may be entertaining and even insightful, do not draw any
serious conclusions from it.

For those seriously interested in the subject, there is an ongoing discussion on the Auditory list about human HF perception, available reliable data (very scarce!) test methodology and possible perceptual mechanism (bone conducted and modulated cochlear excitation is one current theory supported by experiments).

best regards,
Halcyon

PS Do not assume this post to be an argument against high ultrasonic hearing at beyond 22 kHz. Some research data (both psychological and neurological) supports strongly the perception of ultrasonic harmonics even for test subjects who cannot hear single tone ultrasonic test signals. This even appears to influence the enjoyment of musical signals. References available on request (too time consuming to dig them out right now, I don't have them at hand).

I am intesrested, and I request one or two references (as well as the Auditory list URL), if you don't mind

One such reference currently being debated on rec.audio.high-end is:

http://jn.physiology.org/cgi/content/full/83/6/3548

ff123

Thank you for the link. I've been wanting to read this article for a long time. Now I'll be able to speak of it saying less misinformation...

Example, I said that the fact that the EEG that appeared with the high frequencies didn't disappear when they are cut proves that it is not related to them. Wrong : the articles say that the EEG appears AND disappears with the high frequencies, but several tens of seconds after them.

It suggests that the difference might be only audible after several tens of seconds. I might try again some blind tests with longer samples. I'll always remember how the first member of that forum I forgot, that is about home studio, recognized the 16 bits truncated from the 16 bits dithered sample in a blind ABX test : with one listening session per day, in the morning ! Several other golden ears failed before him (one or two other succeeded after him, AFAIR).

People need to quit bashing the SB Live series of sound cards . They make a great choice if you use the SPDIF out for an external reciever. I am pleased with the cards quality, and I cannot tell any difference between the quality of the 44.1khz sweep and the 48khz resampled when using my Kenwood VR-505 reviever (hooked to SPDIF) and AWIA HP-X222 headphones (claimed response to 25khz?) Enought ranting for now...

Sometimes 3 seconds, sometimes 4, in a noisy laptop with the fan kicking in and out once and again, crappy speakers as well... I'm 25, but AFAIK, I'm sure i'll couldn't hear anything above 18KHz. I voted 17KHz.

Bad resampling in these cards occur also using the SPDIF output.

As to differences, It seems that I could tell apart (via ABX) the original the original versus one resampled, played and recorded via SPDIF. Don't have the reference now, but it's here, in the forums.

However, it's strange that you don't hear the effects of bad resampling with the sweep tone, maybe the effect people hear are not due to resampling but due to bad antialias reconstruction filters plus intermodulation in the card?

Yes, same as me. It's very interesting, and seems that some of the things I read/remembered were not accurate about this experiment. It seems that the differences are really audible, and at normal listening levels!!

However, IIRC when the people who did this experiment were asked for more data in order to try to duplicate the experiment, they didn' reply.

Anyway, the study looks quite serious, and is worth some home-testing. The pity is that I don't know of any musical long clips sampled at higher frequencies than 44.1/48 HZ to do testing. Anyone?