[Mike Walker]

Greg Freeman, all sounds (as pointed out) are made up of their base frequency, plus overtones. Remove the overtones ("harmonics") and all that remains is a sine wave. Any other shape, and harmonics are still present. A base tone MINUS all harmonics is ALWAYS a sine wave. Removing harmonics doesn't just "round off the points", it completely "rounds the wave" until it's a....you guessed it, SINE WAVE! Any other "shape" of a waveform is caused BY THE HARMONICS!

Since the 2nd harmonic of a 15khz tone is 30khz, an octave beyond the hearing of the average person, frankly it makes no difference what the "shape" of the wave is...we PERCEIVE it (if at all...remember 15khz is on the edge of the average person's hearing) as a sine wave!

 

[gerG]

Hi Mike.
I believe that most natural sounds are sinusoidal, and I understand the combination of base tones and harmonics (overtones?). However, what we are talking about here is not a natural sound. It is a result of converting D to A, and the result does not have to obey the same physics as a violin string.

Always on the quest for enlightenment, I spoke with an acoustics engineer here at work. He agreed that hearing beyond 15khz is something that most of lost the first time we fired a gun without hearing protection. (He can't even hear crickets anymore!). His take on the issue is that periodic waves can be recorded at 3x rates, converted to the frequency domain, and then converted back at a higher rate. He said that random signals still need the 10x rate. I assumed that he was refering to jazz

I remain suspicious of a system that requires filtering and processing to get back the signal just barely within the range of hearing (but maybe not the range of perception). Going to higher sample rates gives me one less thing to suspect in the audio chain. Since it is now available, seems a shame not to take advantage of it.

You have to wonder where this discussion will be 10 years from now.

I'm out of here for a few days. You guys have fun.

 

[gerG]

before I take off. You said:

"We also have probe microphone curves of the ER-4's measured on various people in the company."

Where the hell do you install the probe? Or do I even want to know?

Do you put it in the other ear and turn the volume way up??

Sorry, this popped into my head last night when I was installing my Etymotics, and the visual image cracked me up

I will admit to trying to fit an ER4S and an ER4P into the same ear for an A/B comparison. Needless to say the results were inconclusive.

 

[Mike Walker]

Greg Freeman writes "I remain suspicious of a system that requires filtering and processing to get back the signal just barely within the range of hearing"

Sorry Greg, but although lps and analog tape can sometimes record sounds higher than 20khz, they do so only at GREATLY reduced level! You cannot, as you can with digital, record an 18khz tone at 0dbfs! (or just "0db" on an analog recorder). You'll get nothing but distortion with lp, and self-erasure with analog tape! When it comes to recording full-on high frequency information, at realistic levels, digital wins by a LANDSLIDE!

As for filtration somehow being evil, I find it amusing that people who most often make this argument are so called "high end audiophiles", who swear by the likes of single ended triode tube amplifiers, which themselves take a NOSEDIVE above about 15-16khz. Talk about filtration! LOL! Peace, Brother!

 

[Gluegun]

Out of curiosity, how does one TEST what frequencies they can hear up to?

 

[thomas]

the best way is to go to an audiologist, they can do the test much more accurately than you can on your own.

If you don't want to, you can try using a test CD (or using somthing like soundforge) with a frequency sweep, and find out when you can't hear it anymore.

There is also a program called NHC tone (can't remember where its from, search headwize when it comes back) that allows you to enter in any frequency and it will play it back. I can hear all the way to the upper limit of that program (soundcard limit at 22khz) but i can only bearly hear bass below 35-40hz... Obciously that test isn't very accurate, but i'm still a bit worried about my hearing in the lower bass frequencies, so i'm getting it checked by a real audiologist...

 

[d_wilson]

Thomas,

We sell a probe microphone system which uses a silicone tube about 4 inches long, and about 1mm OD. It is equalized flat to high frequencies. You place the microphone tube close to the TM (eardrum) and then insert the earphone. If you touch the eardrum while inserting the tube you hear a small thud, and usually it is uncomfortable. One of the slight variables in the measurements at very high frequencies is caused by tube placement.

Listening tests

Everyone has hearing thresholds that are frequency dependent. There is what is called the minnium audible pressure, which is the level at which you humans can hear sounds. Is about 30dB SPL higher at low frequencies and high frequencies. Audiologists use what we call "hearing level" which is the spl level + the minnimum audible pressure. There are a number of papers on this subject. I can post the curve and chart of spl + MAP tomorrow when I get back to work.

Also, you can't count on your speaker system to be flat when you check your hearing (especially your computer speakers). I would trust your sound card to be flatter than your speakers.

When we test microphones we have a very flat control microphone in very close proximity to the test microphone. Using a servo (or compressor) the control microphone measures the speaker level and adjusts the voltage to the speaker to keep it at a steady level. The other computerized method is to do a substraction between the two microphones to get the net result.

Don Wilson
Etymotic Research

 

[grrr223]

I just finished a signals and systems class, and for the first time I understand how all sound that we here is made up of the scaled sum of signals at all of the frequencies we can hear (that may not have been correctly worded, but I think you understand what I'm trying to say). I also learned that by convolving sinc functions with the samples, you can perfectly reproduce a band-limited signal from a sampled version of it that has been sampled at twice the highest frequency present in band-limited signal. Keep in mind, this assumes that the sample values are not discrete (as in a CD), only that the samples are discrete in time.

My question is this:
In a typical CD player, does that convolution with the sinc functions occur, or is 44.1 khz fast enough that the ear can piece everything together just by playing the samples themselves and not trying to recreate the original signal with Fourier transforms?

Does any of this make sense to anyone, I may have to edit this later to clarify a few things if people don't konw what I'm talking about, but I think it is a logical question.

 

[d_wilson]

grrr223

The first rule of electronics is "Don't let the smoke out"