Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Plagued by "inaudible sound" (Read 33005 times) previous topic - next topic
0 Members and 2 Guests are viewing this topic.

Plagued by "inaudible sound"

Reply #25 –
What you are experiencing may be due more to the effects of ATH than hearing frequencies above 20kHz.
Just because other people do not hear the things that you hear does not mean that it's because the sounds are above 20kHz.  The Absolute Threshold of Hearing is dependent on frequency and age. Check out the graph in the Hydrogenaudio Knowledgebase for ATH. Those sounds you hear are probably just too soft for others to hear at that frequency, and your tests at night are helping you to get as close as possible to your ATH.

Plagued by "inaudible sound"

Reply #26 –
Quote from: MasaMuneCyrus on
I'm still interested in why certain LEDs seem to give off noise (my cousin's surge protector is particularly loud, but in that case it could be the surge protector itself, and not the LEDs on it). Whatever the noise comes from, if an LED is incapable of producing it, it surely has to come from the mechanism that turns the LED on, because in the case of my DS Lite and Razor the noise only happens when the LED lights are on, and not when they're off.
  Anything that has AC or square wave in it, in theory is capable of producing sound under the correct circumstances.  Normally, it is chokes, coils, or transformers that vibrate, but certain capacitors, PCB traces, or or even conductors can vibrate and make noise.

Plagued by "inaudible sound"

Reply #27 –
The following paper from the Journal of Neurophysiology might be of interest, although not directly related to the issue sketched by the thread starter:

Oohashi et.al.: "Inaudible High-Frequency Sounds Affect Brain Activity: Hypersonic Effect"

Whether it's resonable to argue that ultrasonic spectral energy content will affect the audible band is an open question. Surely, most sound sources (and receptors) are very non-linear and there are modulation effects in play. This survey of the spectral energy content of music instruments shows many have spectral energy well above 20kHz. But again, if it affects the audible band is an open question. Surely, increasing the bandwidth isn't a bad thing.

Plagued by "inaudible sound"

Reply #28 –
Quote from: MasaMuneCyrus on
In any case, I've listened to generated tones before and they got up to about 22 kHz before the speaker couldn't produce anything higher, and I could still hear it (albeit, just barely). I'd like to find some speakers that can play sound beyond what I can hear, then take it home and try it out at night in complete silence to definitively test my hearing. Unfortunately, I just tried the tone:// generated tones on Foobar on my PC here at home (both headphones and speakers) and it seems to have a cut-off point of about 22 kHz, as well (23 kHz is a lower pitch and I'm reasonably sure that 24 kHz doesn't play at all since 23 kHz is easier to hear than 22 kHz). Perhaps I could try my cousin's new computer with some good headphones this weekend.


It sounds like your sound card is completely unsuited to this kind of test.

Use one that runs natively at 96kHz sampling frequency, and is known to do so cleanly and without adding distortion to high frequency signals under whatever operating system you're using.


FWIW there are useful results up to 24kHz...

Henry, K. R.; and Fast, G. A. (1984).
Ultrahigh-Frequency Auditory Thresholds in Young Adults: Reliable Responses up to 24 kHz with a Quasi-Free-Field Technique.
Audiology, vol. 23, pp. 477-489.

This is with _very_ loud, clean, undistorted sounds. To be detected above 20kHz, the typical thresholds were over 100dB SPL. The lowest found were about 80dB SPL IIRC, and some were over 116dB SPL. You need serious dedicated equipment to deliver this kind of sound!

Cheers,
David.

Plagued by "inaudible sound"

Reply #29 –
Quote
Use one that runs natively at 96kHz sampling frequency, and is known to do so cleanly and without adding distortion to high frequency signals under whatever operating system you're using


This will be the problem you had with the cutoff of 22kHz: since the majority of sound equipment is designed for the standard of 44100 Hz, it can only accurately produce sounds of below half that.
I can honestly say that I occasionally have the same problem with my CRT monitor: leaving it on overnight causes endless irritation. I'm 17, and neither a friend of the same age nor my sister can hear that. Having just experimented in Foobar, I can hear up to about 20.5kHz, though this is in a slightly noisy environment with a poor sound card.
Quote
As far as I know hearing is usually determined by using simple tones, and what is normally accepted (20Hz - 20kHz with possible high frequency drop-off) is very accurate and there is no point debating it.


This is obviously a problem that varies from person to person, and I agree that it isn't exactly true to say that 22kHz is an absolute threshold, as we're all individuals.
Another possibility is that, while we are talking on a very small scale, it could be that the (apparently) very high 40kHz sound emmited by the bulb etc could be causing something else very nearby to oscillate at a different rate, and this would be what you are hearing. I give an example. In our music department, we have an extremely high quality recording pc, with two speakers of fairly good quality. However, in the same room, we have two snare drums, which always resonate with the speakers, no matter what frequency.
While this is obviously on a larger scale, it is a possiblity that this could be happening.

Hope I've been of some use, and feel free to contradict me on anything I've said

Pingue
If at first you don't succeed, so much for hang-gliding

Plagued by "inaudible sound"

Reply #30 –
Quote from: Pingue on
I can honestly say that I occasionally have the same problem with my CRT monitor: leaving it on overnight causes endless irritation.

The problem here is that you actually don't know the frequency of the tone that you are hearing coming from your CRT.

I can hear up to 18kHz (if that) but have no trouble hearing CRTs.

Plagued by "inaudible sound"

Reply #31 –
Theres been threads on these forums about CRTs.  They're actually around 16kHz as I recall, which is why they bother people so much.

Plagued by "inaudible sound"

Reply #32 –
Quote from: Bourne on
The Nyquist Theorem says we can not hear anything above 20 kHz. I barely can hear 16kHz. Human adults above 30 years old have hearing loss they can't even reach that 16 kHz.

WHATEVER!!! But Nyquist was right in a certain practical way... There is a REASON why 22050 was USED. Now ask yourself why Nyquist used this samplerate and not a higher one?  If I can recollect it had to do about what we can hear and the enough safety margin to re-produce analogue sound without a brain to figure out it was digital. Something like that.
 

There were no CD players in Nyquist's day.

Plagued by "inaudible sound"

Reply #33 –
Quote from: Mike Giacomelli on
Theres been threads on these forums about CRTs.  They're actually around 16kHz as I recall, which is why they bother people so much.

PAL line frequency is 15625hz and NTSC is 15750hz. This applies only to TV's though, PC CRT monitors usually have their line frequency well above 40khz.

Plagued by "inaudible sound"

Reply #34 –
Quote from: Bourne on
The Nyquist Theorem says we can not hear anything above 20 kHz. I barely can hear 16kHz. Human adults above 30 years old have hearing loss they can't even reach that 16 kHz.

WHATEVER!!! But Nyquist was right in a certain practical way... There is a REASON why 22050 was USED. Now ask yourself why Nyquist used this samplerate and not a higher one?  If I can recollect it had to do about what we can hear and the enough safety margin to re-produce analogue sound without a brain to figure out it was digital. Something like that.


Nyquist had nothing whatsoever to do with the setting of a 44100 sampling rate.

The Nyquest theorem says nothing whatsoever about what one can hear.

Goodness.

Try Brian C. J. Moore's book on the Psychology of Hearing.

Quote from: jmartis on
The "bug-scaring" devices emit a tone below 10khz I think.
I can hear bats too 
LED's can't emit noise.

Bug-scaring devices often overdrive the units used to radiate the sound in fashions that creates both sub-harmonics and difference tones that are in the audible range. Ditto for some kinds of the old ultrasonic alarms that people used to use for motion detection.

Bats emit clicks (broadband) as well as HF pulses, or some of them do. Some of that energy in the click may very well be well under 20K.

And finally, some kids, at least, can hear well above 20k, but it seems to go away long before adolescence, at least in the modern world.

Then again, it's (*&*( hard to test a 3-year old formally.

Quote from: Mike Giacomelli on
Theres been threads on these forums about CRTs.  They're actually around 16kHz as I recall, which is why they bother people so much.


U.S. Color TV (NTSC) is 15735Hz. Yeah, easily audible.
-----
J. D. (jj) Johnston

Plagued by "inaudible sound"

Reply #35 –
Quote from: jmartis on
Quote from: Mike Giacomelli on

Theres been threads on these forums about CRTs.  They're actually around 16kHz as I recall, which is why they bother people so much.

PAL line frequency is 15625hz and NTSC is 15750hz. This applies only to TV's though, PC CRT monitors usually have their line frequency well above 40khz.

At the conference center where I work, we occasionally use some older 29" NEC CRT monitors as confidence monitors.  They are capable of 1600x1200@60Hz, and we usually run them at either 1024x768@60, or 1280x1024@60.  At either resolution, one of the monitors will sometimes emit a very noticeable high-frequency tone which can easily be heard from several feet away.  I've never had a spectrum analyzer around to actually measure it, but it is definitely lower in frequency than the ~15.75 kHz of an NTSC monitor.

I'm not familiar enough with the inner workings of CRT monitors to be able to determine the exact source of the tone, but whatever the source, it appears that as they age, it is possible for a data monitor to emit audible tones well into the "normal" range of human hearing.
"Not sure what the question is, but the answer is probably no."

Plagued by "inaudible sound"

Reply #36 –
Quote from: mixminus1 on
I'm not familiar enough with the inner workings of CRT monitors to be able to determine the exact source of the tone, but whatever the source, it appears that as they age, it is possible for a data monitor to emit audible tones well into the "normal" range of human hearing.

The tone is emitted by coils that can move as current passes through them. As they age, the glue holding the coils together can soften and intensity of the tone increases, and sub-harmonic tones can be produced too (this is why you sometimes can hear a high-frequency tone from old monitors even if they run at high resolution).

Plagued by "inaudible sound"

Reply #37 –
Thanks for the replies, again. My cousin has a "high-definition" sound card, as well as some decent Creative speakers and some nice headphones. I'll see him this weekend, so I'll have to see what happens over there.

Plagued by "inaudible sound"

Reply #38 –
Quote from: MasaMuneCyrus on
Quote from: AndyH-ha on

The sound might be from some part of the power supply circuit that controls the blinking rather than from the LED itself. I suspect the mechanism is that something is made to vibrate due to being suddenly hit with a surge of power.


That's very possible, as it tends to only be LEDs that are only on when charging, or in my razor's instance, the LED that lights up once it's done charging. Thanks for the info, too.


There are also other electrical components that cause this. I have a diskless system in a silent room based on a thinkpad bottom no display. When I start the system processing heavily I can hear chirping emitted from the TP. I'm guessing it might be a loaded electrical component ie these have energy saving features and processing might dictate the whole processor be used?
That was just a confirmation now I have a weird story.

I have lived in an apartment under 2 different tenants now. I suspect I may snore. I was suspitious 8-9 months back when over 2 weeks I developed a sleep disorder and I never have any problems sleeping. It's like this I'm watching TV in another room and near bedtime get very tired almost nodding off. When I hit the sack I feel tired but the only way I can describe is your eyes want to stay open. At the end of 2 weeks I thought I was going to die I expect it was more than lack of sleep but disorientation too. Unable to prove it, I think that tenant was using a bug or rodent chaser device with the intention of keeping me awake. The disorientation leads to a jumpy feeling almost immediately.
I have not had another problem till 2 nights ago (with new neighbors of several months) I awoke to hear the tenants overhead at 2PM yelling then the rest of the night I mostly lay there. Thats not all my senior cat sleeps on the bed and she sits upright staring in the air. Last night after I figured it might be going the same way I used earplugs which seemed to help but there was still not a deep sleep and the woozy disorientation exists. It is very hard to describe but like a switch. I cannot hear any sound, odd pitches or otherwise. This is a small town and it would not be impossible for the previous tenants secret be shared through friends. Possibly they got the original idea online much like this.

Is it plausible? I guess the only way I can figure it out is to detect the sound like previously described here. If I were snoring it would be nice to know and if a  rodent unit were in misuse like that, wouldn't it be illegal?
Thanks for any ideas.
-John

Plagued by "inaudible sound"

Reply #39 –
Regarding both possible snoring and near-ultrasonic pest control devices (rather than those at 100 kHz) I'd suggest you get or borrow a soundcard that can record at high sampling rates (e.g. 96 kSa/s, which could allow you to record frequencies up to 48 kHz subject to analogue filtering letting that through). Use this soundcard to record the input from a microphone (a small, relatively cheap tie-clip electret microphone could very well be capable of some response at ultrasound frequencies, and if it's loud ultrasound, it should be fairly detectable). At 16-bit 96000 Sa/s mono PCM (such as uncompressed WAV) a recording should occupy about 690 MB per hour (don't use lossy compression as it ought to discard ultrasound). So an overnight recording of 10 hours would occupy 6.9 GB, and 24-hours would occupy 16.6 GB, both easily manageable with modern hard-disk sizes. If you can repeat this over a number of nights you could try to correlate the amount of sleep you get with the presence or absence of ultrasound.

The spectral view (spectrogram) in a program such as Audacity or any spectrogram visualization you might install in foobar2000 (which won't require you to load the whole WAV but will let you skip) should help you identify any strong ultrasound. And listening back to the audible sound at a few times through the recording will tell you if you snore!

There can be other things that might keep someone awake, such as nagging feelings that, for example, one has to complete something by a deadline (such as a tax return one has been putting off), and doubtless many more causes that sleep disorder clinics might be able to identify.
Dynamic – the artist formerly known as DickD

Plagued by "inaudible sound"

Reply #40 –
I jump in at this point with a little test. The original poster, MasaMuneCyrus stated that it is easier to hear 23000hz than 22000hz. I can confirm that.

Although I am not convinced at all that I can hear anything above 20.000, I can still hear something even at 23500Hz (which is almost the highest frequency foobar is able to generate, it's max is I believe 24999), but I can't hear anything at 24000.
The interesting this is, that somewhere between 19kHz and 20Khz the pitch of the sound is not getting any higher, but actually LOWER!!!, i.e. 23000 seems to be lower than 22000hz...
For me, 23000Hz sounds very likely to 3000Hz, and 23990Hz is heard almost as an infra-sound (very deep bass)... Therefore it is clear that I am not hearing 23000Hz but a different tone that is the 'remainder' of something else - more on this later.
BUT!
There is a very steep hearing curve - or call it hearing treshold.
If I set the master volume to MAX, I can clearly and EASILY hear it. Well I can hear something that should be 23000Hz, but it is definitely a much lower frequency. If I lower the volume by 2 'units' in windows, suddenly I can still hear the tone, but with much less strenght. A further decrease of the volume by 1 unit renders 23000Hz clearly inaudible to me.

The conclusion:
1. 23000Hz is easier to hear than 22000Hz on my current HW/SW config, becuase the perceived frequency is lower. This might have been the case also with MasaMuneCyrus.
2. This is obviously caused by imperfection in the loop (HW, SW, human).

To investigate #2, i have created 3 hypotheses.
The reason of the appearingly decreasing perceived frequency must be either:
a) somewhere in the HW. This unlikely though, since:
My notebook features a SoundMax 1986A HD sound chip that boasts 96Khz sample rate (some doubts here, it is possible that it's just 2x48kHz stereo), and I am using Koss Porta Pro headphones that claim to have a frequency response of 15-25000Hz.

b) somewhere in SW. This could be foobar2000, the directsound mixer, or the soundchip's driver itself. Initially I thought it was a bug with the directsound mixer. To test this I have replaced the directsound output with ASIO (showing max. output frequency of 48kHz - that's way I believe 96kHz in the specs is just 2x48kHz). However, the result was still the same. Since I believe foobar as a quality soft (I don't have winamp to test, and Audacity can't generate a tone higher than 20kHz), the sound driver could be the only possible sinner in this scenario. Is there a way to test this?

c) my ears perceive very high frequencies the way they do, i.e. as a lower frequency... Again, this is unlikely. OR NOT????

What do you think of this?
Matyas

PS: Audacity has real problems playing back two high pitched sine tracks. One with a tone of 19000Hz and the other with a tone of 20000Hz when the samplerate is 44100Hz...

Plagued by "inaudible sound"

Reply #41 –
Quote from: Matyas on
The conclusion:
1. 23000Hz is easier to hear than 22000Hz on my current HW/SW config, becuase the perceived frequency is lower. This might have been the case also with MasaMuneCyrus.
2. This is obviously caused by imperfection in the loop (HW, SW, human).



3) Something in the system is aliasing things down.
-----
J. D. (jj) Johnston

Plagued by "inaudible sound"

Reply #42 –
Quote from: Matyas on
For me, 23000Hz sounds very likely to 3000Hz, and 23990Hz is heard almost as an infra-sound (very deep bass)... Therefore it is clear that I am not hearing 23000Hz but a different tone that is the 'remainder' of something else - more on this later.


I'm pretty sure Woodinville is correct, that this is definitely aliasing (Wikipedia link) or something a little more complex that involves aliasing.

The weird thing is that for a 23990 Hz sinusoidal wave to create an alias at 10 Hz (infra-sound very deep bass), the sampling frequency (not the Nyquist frequency) would have to be 24 kHz* (which implies a Nyquist Frequency of only 12 kHz), and not the usual 48 kHz sampling frequency that is used by DirectSound mixers and Soundblaster Audigy soundcards with known problems.

(*Actually, any integer division of 24 kHz sample freq would also give 10 Hz, such as 12kHz or 8 kHz)

A 48 kHz sampling frequency would give a true 23990 Hz wave. Only looking at the sampling points it looks a lot like a 24000 Hz wave (Nyquist frequency) that is amplitude modulated by 10 Hz (i.e. alternating peaks and troughs in successive samples, whose apparent amplitude goes up and down 10 times per second). An ideal reconstruction filter would, however, perfectly recover the analogue 23990 Hz wave with no appearance of amplitude modulation. It's all down to the eye assuming that the sample points fall on the peaks of the waveform.

You can see the equivalent effect using The Aliasing demo Java Applet which simulates 8000 Hz sampling rate audio. If you put in 7800 Hz, you'll get a 200 Hz alias frequency. If you put in 23800 Hz, you'll get the same 200Hz alias, for that matter. If you use 3800 Hz you get no alias, but looking at the sample points only, you notice what appears a lot like 200 Hz amplitude modulation.

With a poor reconstruction filter it might well be that this AM effect is reproduced on the analogue output signal (this effect is actually a superposition of 23800 Hz and 24200 Hz sinusoids assuming a 48000 Hz sampling rate and poor filtering). Some non-linear components in your signal chain might then even act as mixers and produce audible difference frequencies (as well as inaudible sum frequencies) from the superposition.

The point is that if you're anywhere near the Nyquist frequency you're relying on your soundcard having an exceptionally good reconstruction filter to be able to generate the pure tone you intended. That's why a true 96 kHz sampling rate with no known problems/issues is practically essential in performing pure-tone hearing tests, so that the 20-25 kHz test tones are well below the 48 kHz Nyquist frequency.

This is where high amplitude pure-tone testing differs from the sort of low-pass filter tests people carry out on real music and why it needs soundcards that are significantly over-specified beyond the requirements for transparent music reproduction.
Dynamic – the artist formerly known as DickD

Plagued by "inaudible sound"

Reply #43 –
Quote from: Dynamic on
Regarding both possible snoring and near-ultrasonic pest control devices (rather than those at 100 kHz) I'd suggest you get or borrow a soundcard that can record at high sampling rates (e.g. 96 kSa/s, which could allow you to record frequencies up to 48 kHz subject to analogue filtering letting that through). Use this soundcard to record the input from a microphone

Thanks for all the info. I'll have to research an appropriate card and mic. Any favorites? I am really interested
in this beyond my problem so I don't mind buying something reasonable.
I tried with the cards I do have but they only seem to get to 15-16.
As for my taxes they were done but you never know it might be something else.

Plagued by "inaudible sound"

Reply #44 –
Quote from: Dynamic on
The weird thing is that for a 23990 Hz sinusoidal wave to create an alias at 10 Hz (infra-sound very deep bass), the sampling frequency (not the Nyquist frequency) would have to be 24 kHz* (which implies a Nyquist Frequency of only 12 kHz), and not the usual 48 kHz sampling frequency that is used by DirectSound mixers and Soundblaster Audigy soundcards with known problems....



Or some kind of intermodulation artifact. I have seen artifacts at fs/2+-x before that came from distortion products where some kind of clock leakage or other got involved. I did not have the opportunity to track down where it came from, in fact the guy I was working with looked at it for about 30 seconds, and frisbee'ed the card into the wastebasket.

But I've seen that kind of distortion. No, it's not directly alaising. Duno what it is.
-----
J. D. (jj) Johnston

Plagued by "inaudible sound"

Reply #45 –
The M-Audio Audiophile 2496 is supposed to be very good according to posters I trust (e.g. 2Bdecided), and rather inexpensive because it's only stereo, not surround-sound.

Many electret microphones will mention a frequency response (usually where the response drops by 3 dB from the peak) on the packaging, and for tie-clips, top-end response of around 20 kHz isn't uncommon. Hopefully such microphones would still respond enough to pick up loud signals in the few kHz above that.
Dynamic – the artist formerly known as DickD

Plagued by "inaudible sound"

Reply #46 –
I tried the goldwave and i can barely listen to 21.00k sine wave

Plagued by "inaudible sound"

Reply #47 –
I'd expect the high-frequency audio performance of most computer audio circuitry to fall anywhere between awful (full of distortion, harmonics, noise) and non-existent; as such I propose the following experiment:

Find a signal/function generator that is capable of output up to and past the audio frequency range.  This should be relatively easy, as even cheap-o multipurpose function generators are capable into tens of MHz -- well above what any of us will be able to hear!

Hook up the output to a decent pair of headphones (check their frequency response specs, you want to find one that doesn't have too sharp a roll-off in the treble, but not too fancy should you accidentally fry them with too much power -- don't blow up your Sennheiser 580s!  ).

Start with a definitely-audible frequency (e.g. 500 Hz), and find an amplitude (signal voltage) that produces a loud but comfortable volume in your headphones.  Then increase frequency, and see where your ears cut out.  Simple!

Plagued by "inaudible sound"

Reply #48 –
As for the upper range of human hearing.. who knows. I've heard it can go up to ~24 kHz.

You're likely hearing the power supply for the fluorescent lights. The LEDs that you can hear are likely using PWM to control the brightness of the LED. Basically the power supply rapidly pulses the LED on and off, and if this pulsing frequency is in the audible range (< 20 kHz) it can be audible - it's not the LED that's causing it though. Newer constant current drivers should be quiet.

I've read that even older adults with limited hearing above ~15 kHz can still hear and respond to subliminal messages embedded in audio ("silent sound") above 15 kHz, even if they are not conscious of it.

Plagued by "inaudible sound"

Reply #49 –
WOW!!! I've learned a lot reading this thread.
That frequency thing in Foobar was pretty cool 

But I got suspicious when I heart frequency's through the headphones up to 23999 Hz.
And when some of the frequency's I tested changed and played lower than the previous I guessed something was not right.

So ... I made MP3's of them (-b 320 -k) and got 48 kHz mono files. Expected 44,1 and stereo but decided to test the just the same. And it worked 
Tested on a Harman Kardon AVR 430 (7x75W) + JBL Northridge E-100 (2x250W) and JBL E-250P subwoofer (350W).

Interesting to see the speaker cones at 2 Hz. Just pulsating slowly back and forth.
Started hearing something at 6 Hz. The amp. seemed to be working hard. Heard some fans (?) starting to work.

Made files @ 2-32 Hz in 2 Hz increments, then doubled the Hz each step up to 16.000 Hz. Then 1000 Hz increments up to 20.000 Hz and some above that. I needn't have bothered. I heard 16 kHz but not 17 kHz.

So I'm guessing my limit is around 16.500 Hz      or  or 
FYI I'm 32 years young