Poll: Can you hear sound over 20kHz? - Page 17

 

OK, I will keep reading..., soo you're saying if you zoom in enough on a triangle or sawtooth waveform they're actually all sinusoidal right?  In nature too?  What about something like a Commodore=64 or Casio sound chip?

 

 

 

It's not the fundamental note, it's the overtones etc., which is why 44.1kHz sounds a lot better than 32kHz, I can easily hear it with vocals on my useless laptop speakers, despite the fact no one is singing notes above 16kHz. =p

Edited by kiteki - 5/10/12 at 7:53pm

The core frequencies of music don't go anywhere near 16kHz. In fact, there have been studies where people were asked to listen to music with everything above 10kHz chopped off and most people said it sounded fine. 20kHz is about three octaves of upper harmonics... Overkill.

A person whose hearing only goes up to 15kHz is only missing third level harmonics on cymbal crashes. Everything else will sound fine.

i cant hear past 16khz. none of my friends can hear past 17khz. i know a classically trained pianist who has babied the crap out of her ears and she cant hear past 17khz

I'm confused, I thought 44.1khz was the amount of cycles recorded per second or something. I had no idea it meant the sound went up to 44khz. If people can't hear above circa 20khz what would be the point of having higher frequencies that are perceivable? 

44.1 only goes up to a little past 20kHz. Two different measures.

What xnor means is that there is no straight lines in audio because of its nature as a band limited signal, thus, there is no true straight square waves or triangle waves in audio either. A true square wave is the sum of an - infinite - number of sine waves as described but the following sum:

And since audio is band limited, the sum is stopped at some point, and all you have is an approximation of a square wave, thus the Gibbs phenomenon.

I would suggest reading on Fourier decomposition and harmonic analysis to understand the issue, the wiki articles are not bad, but can be challenging for the layperson.

 

Link it.

 

 

 

44.1kHz = the sound goes up to 22.05kHz

 

In the real world there's a lot of sound above 22.05kHz which creates a lot of audible noise in the recording, so you have to remove the noise with filters.

 

From the air in front of the mouth of a singer to the air in front of your speaker, it should be a pure process with as few filters as possible.

 

At extremely loud levels people can hear sound up to ~25kHz, at what level we can perceive sound is inconclusive, but there are theories that insect repellors or cleaning devices can cause stress with their ultrasonic frequencies, and likewise that 24bit / 192kHz audio sounds better than CD quality, of which there is a cult following.

It come from the theorem of Nyquist Shannon, to reconstruct a signal band limited to 20 kHz (sound), you have to sample it at double its frequency, in this case 40 kHz.
For CDs, the sample rate is 44.1 kHz, means that the reconstructed sound can have frequencies up to 22.05 kHz.

 

Ok thx that makes sense!  If a square wave is an infinite number of sine waves...

 

However it seems like digital processing draws straight lines, such as in this example - http://www.head-fi.org/t/571259/hi-rez-another-myth-exploded/15#post_7748362

I am skeptical at that claim, given a sampled 22.05 kHz sine wave and a 44.1 kHz, it's will be reconstructed into a 22.05 kHz sine wave; if it were reconstructed into a 22.05 kHz triangle wave, it would require an infinity of harmonics.

192/2 = 96khz? 

Man that is a lot of frequencies I wouldn't even be able to hear. Wonder what difference that makes. 

Indeed.

And considering that no microphone is linear up to 20 kHz.... yeah it's pretty useless.

a) Yes
b) This guys' drawing don't make sense, not even the analogue one.

 

You don't necessarily need to hear it.

 

"Hosoi H, Imaizumi S, Sakaguchi T, et al. Activation of the auditory cortex by ultrasound Lancet 351:496-497, 1998.

Imaizumi S, Hosoi H, Sakaguchi T, et al. Ultrasound activates the auditory cortex of profoundly deaf subjects. Neuroreport 12(3):583-586, 2001."

 

I would have thought reverb is typically a linear time invariant effect (i.e. convolution with an impulse response, which does not produce any aliasing), but even if time varying delays and/or slight distortion are added to increase the complexity, it is not clear why it would need to be more than 20 times slower than real time on currently available hardware to avoid aliasing; this sounds like snake oil.