I am a bit puzzled, because that statement is wrong and we have discussed the "hypersonic topic" a couple of times during the last years in this thread/forum.
It starts with Oohashi´s articles and there is some more research from other people as well.
Results are inconclusive so far but it way more than personal anecdotal tests.
No USB here (not that there's anything wrong with that...). M-Audio 192 PCI card, Echo Audiofire 2 Firewire card.
In 2008 I did Fresno to Portland, OR. Beautiful! Yes, the train was late in Sacramento, but that was fortunate because it put us at the foot of Mount Shasta at sunrise. What a glorious sight. One of the best train trips I've ever taken. It's a spectacular route. Yes, it gets tiring being in the same seat that long, but at least you can stroll around the train, unlike the flying sardine tins.
Ed Simon - looking forward to seeing your further noise measurements, including batteries.
In my silly tests of different conductors- copper, steel, potatoes, bananas, water, wine, mud - I've been picking up a lot of mains noise right out of the air. Interested to see what you are getting with better methods.
In my silly tests of different conductors- copper, steel, potatoes, bananas, water, wine, mud - I've been picking up a lot of mains noise right out of the air. Interested to see what you are getting with better methods.
Great idea. I look forward to your articles on how to do it.
If you want to try an actual experiment try driving a tweeter with 30kHz. With a bit of level you just may find although you don't hear a tone you can tell when it is on or off.
Every time I demonstrate this folks insist it must be subharmonics they hear.
Now when I hook a compression driver to a test oscillator, as I sweep it down different folks complain at different frequencies. Always below 18 kHz.
Simple experiments almost anyone can do.
Demian,
A real issue with any digital system is EMI leakage into the rest of the system. It is clear to me tweaking a CD player with an EMI resistant but slightly higher distortion opamp does produce more pleasant results.
Gerhard,
If you are doing a serious survey of batteries, I will not duplicate that. The issue with noise vs current seems to be the noise decreases with current draw until the battery begins to heat. This varies a great deal but seems to be around ten percent of capacity.
Note typed on my cell phone so all spelling corrections I missed don't kount.
ES
OK, so there is no perceptual argument to your numbers, but it is linked to what we can do today. Ten years ago you would have insisted on a minimum of 96kHz sampling to catch all we can hear. Today it's 196k. Tomorrow maybe 512k?
Wish my hearing would evolve that fast! 😀
BTW I agree that supersonic tones can cause brain activity. I think we disagree on how high up we need to go to reproduce the audio experience.
jan
I have mostly seen enthusiastic interpretations of Oohashi's tests. They were not DBT tests of music filtered/unfiltered. There have been several carefully controlled listening experiments with 20k brickwall in/out with null results that are repeatable.
I disgree with you.
BTW I think it was one of Greiner's papers that pointed out the extreme difficulty of testing this hypothesis with any real loudspeaker due to it's own distortion.
EDIT - I see Oohashi himself is now exploring non-auditory sources for the "hyper-sonic" effect.
Last edited:
Both Manfred Schroeder and Rupert Neve reported examples of 5 degrees of phase shift at 20 kHz being heard. Now I have been unable to repeat that. But 30 degrees at 10 kHz does show up. You are welcome to try this yourself. Neve spotted a defective transformer. Schroeder used DSP waveforms that had the phase shift but kept the amplitude constant.
So it seems a reasonable design goal to have less than 5 degrees of phase shift at 20 kHz. Now how you want to do that is up to you.
Note that if you record it at a higher data rate you can use DSP to produce a file that will met those requirements at a lower data depth and rate. Doing it with analog filtering borders on the silly.
these were diotic phase shift?
we clearly can detect 10 us, possibly less in interaural test with headphones, special clicks of enveloped few kHz test signals, shifted in time between R/L ears
but phase doesn't seem to be directly encoded above 4-5 kHz - simply because most nerves maximum firing rate saturates
we clearly can detect 10 us, possibly less in interaural test with headphones, special clicks of enveloped few kHz test signals, shifted in time between R/L ears
but phase doesn't seem to be directly encoded above 4-5 kHz - simply because most nerves maximum firing rate saturates
DF96
I can not visualise such a case. (It is my poor english, sorry).
Do you mean "when sampling signals having frequency components above the Nyquist frequency for the chosen sampling rate ?
Thank you for that wise post !
George
Better get your head vices out for the mics too. How many stereo pairs do you think are set up that carefully. How about the musicians moving around, what, fractions of a mm.
I'm not sure what you are asking. My point is that to avoid aliasing you need no signal components above the Nyquist limit. There are two ways to achieve this:gpapag said:
1. use an antialias filter - as in normal digital audio
2. ensure (or assume?) that no such frequency components are present in the analogue source signal - as used in some data sampling (which is where the advice about 5x or 10x comes from)
There is a third way:
3. decide that a little aliasing won't do any harm - often combined with 2. above?
If you sample a signal at 10x the highest frequency you can fairly easily visualise that you have a good digital representation of the analogue signal. Sometimes that is sufficient - you will have images but they will be small enough to ignore.
If you sample at 2.2x the highest frequency you still have a good digital representation of the analogue signal, but you need to use a good reconstruction filter to get the analogue back as without this the images will be nearly as big as the wanted signal.
It's pretty good, I have one running, but be aware that it is not a sound card in the traditional sense - it only runs the propriety software.
jan
I have seen reports that the transition when phase is changed at high frequencies is audible but the actual phase relationship would be a different issue I think. There are reports that phase is not audible. However since keeping it intact is not an ovewhelming technical challenge why not?
Maintaining it in the analog domain is not hard if you can work with wide bandwidth. ADC's like the AKM can maintain .05 dB frequency response to 80 KHz at 192 KHz. Doing that may incur some significant phase stuff at the top of the band. Mostly evident in ringing on transients. The waveform fidelity below seems quite good. Perhaps the ringing is the audible effect not the phase shift?
- Status
- Not open for further replies.