Subject: Re: Frequency response was Re: [linux-audio-user] Audiophile CD's
From: Jason (hormonex_AT_yankthechain.com)
Date: Tue Jan 29 2002 - 00:21:43 EET
On Mon, 28 Jan 2002, Paul Winkler wrote:
> On Mon, Jan 28, 2002 at 03:16:04AM -0500, Jason wrote:
> > True, but I'm not convinced that that necessarily means that everything we
> > hear is heard that way because our brains are constantly summing a bunch
> > of sine waves.
>
> That's actually surprisingly close to how it works, IIRC. The ear is
> sort of like a massive parallel bank of very narrow-band filters
> feeding amplitude sensors. FFT, in other words.
>
> > However, I can tell the difference between a 16k sine wave and a 16k
> > square wave, and I can honestly say that the 16k square wave sounds like a
> > square wave, and not simply "different" from the sine wave- It's a pretty
> > easy test to execute, I suggest everyone
> > interested sampling frequencies conduct it at several different sampling
> > rates, it was a bit of a revelation for me- Which means
> > that either a.) you can hear much higher frequencies than 20kHz, and your
> > brain simply ignores them when they aren't associated with a lower
> > fundamental frequency (frankly, that seems a bit unlikely) or b.) The
> > brain doesn't process timbre by performing lightning quick FFT on the
> > input
> > from the audio nerves.
>
> Well, this is very interesting because b) contradicts everything I've
> read about how hearing works. I would very much like to experience
> this for myself, it would change my understanding quite a lot. So I'd
> like to ask you a bit about how you executed this test.
>
> * Did you do a blind test? i.e. was there any way for you to know
> which signals were which?
It was a blind test over several different frequency bands using a high
quality analog tone generato through a neve channel strip into high
quality KRK monitors. A friend and I took turns playing different
wave forms for each other and had the other write down whether they
thought it was a sine a square or a sawtooth wave and then compared our
results. It was fairly unscientific, because we were really only doing it
to learn the distinct sounds of teh different wave shapes and the quality
of different frequency bands, not trying to
prove or disprove anything about psychoacoustics. For both of us, once
we'd become familiar with each waveforms timbre, got results better tahn
would be predicted by change for assigning qualities to the upper
frequency signals. I don't have the scratch paper, but the accuracy was
somewhere near 75-85% correct for 20 signals at greater than 10kHz.
I'd always just taken it for granted, but now that there seems to be some
controversy, I'd be interested in hearing about other people trying a
similar experiment. Maybe I was just deluding my self and tuning into some
imperfection in the particular playback mechanism, that certainly is a
possibilility I'm willing to consider. However, I have heard other
engineers report similar experiences.
>
> * How did you generate and play back the 16 kHz square wave? Did you
> use an analog tone generator, or a digital system with extended
> response? If digital, what did you use to generate the test signal,
> and what did you do about aliasing?
like I said, it was an entirely analog signal path.
> * Do you know for sure that the playback system doesn't produce
> distortion below 20 kHz when faced with harmonic content at 32 kHz and
> above? How can this be verified? I don't know how to do this on my own
> system...
I don't know this for sure, and it certainly is a possibility that either
the amplifier(and exactly which model it was, I can not recall) or the
speakers were introducing some distortion into the signal. However, if it
were distortion, I would think that it would be difficult to tell teh
difference between a sine wave and a square wave, and both my friend and I
were both able to do this at a rate better tahn that would be predicted by
chance. That I'm very sure of, since that was actually teh reason we were
doing the exercise.
I could very well be completely off base here, I'm no psychoacoustician,
but it seems to me that given the model of hearing that folks have been
presenting, that the brain could very well be capable of detecting higher
partials than the traditional 20k assumed limit when they are present with
a closely related lower fundamental frequency.
I'm going to have to do some more reading...
Interesting discussion, though.
-- YankTheChain.com - You can pretend we're not here. That's what I do.,
This archive was generated by hypermail 2b28 : Tue Jan 29 2002 - 00:11:54 EET