Audio Myths Workshop

[aka_bip]

I already have this one
http://controversy.wearscience.com/design/devil/

[Meffy]

@bip: They're all nice designs. Had I the money I'd buy lots of those, but even more of the "SCIENCE!" series.

[Gamma-UT]

Oh, I'm getting me some of these T-shirts. Hooray, the thread's been useful

I'm probably going to regret posting this link but this is from a relatively recent study of ultrasonic noise.

http://www.jstage.jst.go.jp/article/ast ... 2/_article

Four out of the 15 people tested could hear tones above 20kHz. Before anyone starts getting excited and claiming the 22kHz limit assumed by CD audio is BS, take a careful look at the scale to the left of the graphs on p17. I've posted one of the graphs here to save a lot of downloading. This was a test of intense ultrasonic noise: they used 'supertweeters' to deliver 85dB SPL or so at 16kHz and up. 85dB is the limit proposed by the EU for iPods etc.

Even the people with the best hearing in this study could only just detect the sound at these levels. Consider what these people might be able to hear in a full programme and what effect, if any, the -40dB or whatever level you get at this frequency has on their perception of it.



Hearing in terms of both frequency and loudness is logarithmic so not exactly linear but, as it's a predictable curve, hardly non-linear either. It depends on whether your definition of linear means "straight line" or "line with no discontinuities".

[Z1202]

If you have evidence, point me to it.

If you really care for the evidence, then I'd suggest you try to get more info about the detail of the experiment, where people claimed hearing things above 20kHz. If the experiment was scientifically correct, this might be the very evidence you're asking for. There might be some other experiments (I heard of some, but that's unreliable info). Otherwise I probably should assume that you care only of proving me wrong just for the sake of it

[Z1202]

Hearing in terms of both frequency and loudness is logarithmic so not exactly linear but, as it's a predictable curve, hardly non-linear either. It depends on whether your definition of linear means "straight line" or "line with no discontinuities".

There seems to be a constant misreading of the term "linear" which I used. Sorry for the ambiguity, but this is the official term.

As regards to registering the tones in the above experiment, this again doesn't prove anything. Basically it's like this:

- if someone ever proves that human hearing is linear, then anything above 20kHz doesn't matter
- otherwise it's almost impossible to obtain a real proof of the above, and the question will stay open

[aka_bip]

I didn't make any statement, whose correctness depends on the presence of the said teapot. On the other hand, the claim that anything above 20kHz doesn't make any difference, requires either the proof of the hearing linearity, or a more complicated proof for the nonlinear case.

So, yes, the question is open. Can we therefore agree, that the statement about people imagining things when they claim to hear things above 20kHz is open too?

Those claims are not equal, just like my claim that there is an orbiting teapot is less reasonable than your claim that there isn't, despite there being no evidence either way.
If an inaudible signal is combined with another inaudible signal, it is reasonable to conclude that the resulting signal should be inauble, barring evidence to the contrary.

[aka_bip]

Just to clarify, I never claimed that no one can hear above 20kHz. I'm sure there are some outliers that can. I claimed that you cannot make an audible signal from a linear combination of inaudible signals.

[Z1202]

If an inaudible signal is combined with another inaudible signal, it is reasonable to conclude that the resulting signal should be inauble, barring evidence to the contrary.

It is reasonable to assume that, but keeping your mind open to counterexamples should they occur. I have an impression that you are rather willing to discard anything which might be such evidence without a closer look.

Anyway, my point was that the "proof" is not a proof. Can we agree to that?

[Z1202]

Just to clarify, I never claimed that no one can hear above 20kHz. I'm sure there are some outliers that can. I claimed that you cannot make an audible signal from a linear combination of inaudible signals.

It is reasonable to assume (at least for a good majority of cases), but I'd not take it for a hard-proven fact.

[aka_bip]

Just to clarify, I never claimed that no one can hear above 20kHz. I'm sure there are some outliers that can. I claimed that you cannot make an audible signal from a linear combination of inaudible signals.

It is reasonable to assume (at least for a good majority of cases), but I'd not take it for a hard-proven fact.

Of course, but you could say that about most anything. As an aspiring scientist I don't state anything without unspoken caveats: reasonable interpretation of available data, unless new evidence strongly suggests otherwise, etc, etc.

[Nokenoku]

Hearing in terms of both frequency and loudness is logarithmic so not exactly linear but, as it's a predictable curve, hardly non-linear either. It depends on whether your definition of linear means "straight line" or "line with no discontinuities".

There seems to be a constant misreading of the term "linear" which I used. Sorry for the ambiguity, but this is the official term.

It'd be much more correct, if you'd just say:
"I assume, that tones above 20kHz can lead to distortions in a human ear, which produce frequencies below 20kHz."

Let's be honest. If I really push the volume I can hear tones around 20kHz (assuming my headphones and speakers do not both distort at those frequencies in a way, which'd make my experience here flawed).
But those have a) to be very very loud (if I'd do the same with a 10kHz tone, it'd really really hurt), and b) provide no musical value anyways.

[deaf dunderkwac]

...
Let's be honest. If I really push the volume I can hear tones around 20kHz (assuming my headphones and speakers do not both distort at those frequencies in a way, which'd make my experience here flawed).
But those have a) to be very very loud (if I'd do the same with a 10kHz tone, it'd really really hurt), and b) provide no musical value anyways.

you forgot about triangle solos

[ZeroNeuro]

I do all my recording and etc. at 24 bits, 96 khz. Not because I think there is a huge difference in the sound, but mainly because my Apogee audio interface uses a ADC/DAC converter chip, Cirrus Logic CS 4272, that performs better at 96khz. The general latency at 96khz is under 3ms whereas if I go to 41.1k or 48k, it is about 7 to 8ms. I know this is not much even then, but better is better.

I personally cannot hear anything above 22 khz. I've tested myself with an audio file. SIlence. I could see the meter all lit up, but could not hear a thing. Could be my hearing, I am 35 years old, after all.

For the most part I agree with the statements made in the Myth seminar. My reasons for using 96khz is based solely on performance, more than audio quality. I'll use 32 bit float if I need a software VST instrument, and I've been playing with upscaling all tracks to 32 bit for summing. That seems to get around the slight thinness that digital summing can have under certain situations. But I'd need to test that one out more

-ZN

[tony tony chopper]

- otherwise it's almost impossible to obtain a real proof of the above

ok so back to the original subject, CDs would be inferior because they can't reproduce something, but it would be almost impossible to prove it? mmh..

Mmh it should be extremely easy to prove. What would be hard would be to explain WHY it's the case that ultrasonics can end up audible, but not the fact itself.