Record high pitched sound (long post, but please help)

Hello,

I’m fairly new to Audacity and quite the noob, but I think I’ve managed to learn the basics. I downloaded the program in order to record a high frequency sound that comes from the output of my laptop and into my earbuds.

I tried to have the problem fixed, but the guys at the repair shop couldn’t hear the sound. This is because the sound is very faint and high pitched. It was supposed to be a free warranty repair, but since they deny that there’s any fault, I’m left with the bill for diagnostics and shipping costs Personally I hear the sound clearly and have to listen to a tinnitus type sound whenever I watch movies or listen to music.

---

So I need to prove that this sound actually exists. I was thinking that I could record it and boost the volume, so that’s what I tried. I connected a male-to-male jack from the output to the input and recorded the sound with Audacity. I successfully captured the sound, however there’s a lot of static noise when I turn up the volume, which masks the sound. The static isn’t present/audible normally.

I tried the frequency analysis to visually prove the sound’s existence, and there are spikes at around 7 kHz and 14 kHz. I used the notch filter to remove these frequencies, and sure enough the sound disappeared. However I’m not sure the results are conclusive, because my other computer also have spikes at 7 and 14 kHz, but that one has no high pitched sound.

I also tried to isolate the sound by using the notch filter (removing all but the 7 and 14 kHz frequencies), but it seems like when I remove the frequencies next to 7 and 14 kHz, these frequencies weaken as well.

---

These are the results of my experiments:

(Turn down your volume )

Sound clip 1 (with high frequency sound):
https://clyp.it/eqmcu3r0?token=e907b783636396017f01d65ec48dad22
The sound can be difficult to hear through the static, but it’s there.

Sound clip 2 (same recording, but without 7 and 14 kHz frequencies):
https://clyp.it/trqghe5k?token=41cc9e325a860a11d5c16ac6cf9f9279
Only static.

Screenshots:
http://imgur.com/a/lxQhF
The first six pictures are previous recordings for which I have no corresponding sound clips. The next 4+4 pictures match sound clips 1 and 2 respectively. Unlike the first six recordings, these are boosted beyond just turning the volume up in Audacity (which is why they’re so fat). I don’t really know how it happened, it wasn’t intentional, but it occured when I exported and saved the files and then dragged them back to Audacity.

---

So is there a way to record the high frequency sound without static noise? Is there a way to isolate the sound / remove the static noise? Is there a way to visualize the sound in a conclusive way?

And btw, why are the frequency analyses in negative dB?

Sorry for the long read, but I really hope someone can help Additionally, I hope this is the appropriate forum.

Cheers,
Pahans

Please post a short sample (just 4 or 5 seconds) in WAV format to the forum.
The sample should be the raw recording with no additional processing at all. Just “record, select, export selection”.
See here for details about posting an audio sample: https://forum.audacityteam.org/t/how-to-post-an-audio-sample/29851/1

If you have “Y” jack cable, try to record as you did, but with the headphones attached too.

Due to the higher impedance of the input, the recording shows static and noise. If you put headphones in parallel, this could go away.

And maybe the input is a microphone input, in which case it is too sensitive and Windows might try to “enhance” the sound.

Spectrogram view …

The constant ~7kHz tone is on both channels , but the ~14kHz harmonic is only on the right channel .
[ NB: 14kHz is too high-pitched for most people to hear, no matter how loud it is.
7kHz , if loud enough, will be audible by most and sound like tinnitus ].

If you play the track at half-speed the strong peak @ ~14kHz on RHS becomes a loud ~7kHz whine.

If both computers are in the same location possibly both are picking up the same interference [EMI].
Alternatively if your both of your computers been upgraded to Windows 10, maybe you need to update audio-driver-software compatible with Windows 10 , or run the old audio-driver in “Windows compatibility mode”.

Hi guys!

I’m sorry that I haven’t been following up on my own thread - I’ve been out of town (and quite busy) for a while. But I appreciate your responses, they’ve been very useful, so thank you!

Here you go The recording titles are pretty self-explanatory; four of them are from my bad pc (included in this post), the others are from my other pc for comparison (I’ll add them in the next post (upload restrictions)). For some reason, the 7 and 14 kHz spikes in my control pc only occur occasionally As you can see, they’re only present in one of the recordings, while in the past there have been more. I have no idea why, but it might have to do with the hard drive working.

I’ve made recordings using both Wasapi, MME and DirectSound, although I’ve got no clue what the differences and applications are I just thought the more information the better.

---

Do you mean a 2 male + 1 female cable? I didn’t have that, but I had a 1 male + 2 female cable. I connected a separate 2 male cable to one of the female ends (effectively making it a 2 m + 1 f) and then I plugged it into the input/output with headphones attached. However, it didn’t seem to make much of a difference to the recording =/ Does it matter whether I put the spliced end (with the headphones) into the input or the output?

---

Is that one of my pictures? I can’t see the 14 kHz line in any of my pictures on Imgur (that’s the line with no arrow, right?). How did you produce that?

I have to disagree with you there. From what I’ve learnt, humans can hear up to around 20 kHz, and in some cases even more if you’re young enough. Personally I tried this https://www.youtube.com/watch?v=VxcbppCX6Rk test on Youtube, and I can hear the 19 kHz sound perfectly fine if it’s loud enough. I ran the clip through Audacity as well, so I can confirm that the frequencies are accurate.

Thank you for this, that was very helpful! The high pitched sound became much louder relative to the static when I slowed the recording down.

I think I can rule out interference. For some reason, my old computer doesn’t always have the spikes (like I wrote above), contrary to what I believed in my first post. Also I’ve done recordings on the problem computer outside the building; same story. Not to mention that it didn’t use to have a high pitched sound before.

Thanks for the tip, but I don’t want it fixed, because I need to prove that there’s a fault in order to avoid the repair bill

Recordings from control pc.

I’m going a little off-topic here, but it’s still relevant …
Human hearing range is usually quoted as 20 Hz to 20 kHz. There have been no cases (in proper scientific / medical tests) of anyone being able to hear above 20 kHz. Young people are sometimes able to detect sound up to about 20 kHz, but it’s not quite “hearing” in the sense of hearing “normal range” sounds. 20 kHz is the very top of what anyone can detect: “yes there’s something there - a sort of sensation in my head”

Running hearing tests on consumer equipment will often be misleading. When very high frequencies are played at high volume through speakers or headphones, it will often create resonant harmonics at lower frequencies than the actual signal - for example, playing a 19 kHz sound loudly through normal headphones could produce frequencies well below 19 kHz as well as the actual 19 kHz tone. To test properly you need to use proper medical equipment to be sure that isn’t happening.

14 kHz is audible to some people, but for someone with high frequency hearing loss it may not be. Very high frequencies tend to be more difficult to hear when there is other high frequency noise present.

The default spectrogram on Audacity only shows up-to 8000Hz , but you can zoom-out to show the full frequency range , ( or change the maximum frequency in Audacity spectrogram preferences )

That YouTube [v=VxcbppCX6Rk], even when played at top 1080p “HD” quality , does not contain any audio above 16kHz …

The YouTuber has underestimated the limit the YouTube transcoding puts on the audio.

Please do. It’s an interesting topic.

From Wikipedia: “Under ideal laboratory conditions, humans can hear sound as low as 12 Hz[6] and as high as 28 kHz”. Hearing range - Wikipedia
(Sorry, I can’t find out how to hide URLs in words.)
Source material:
Hearing thresholds for pure tones above 16kHz | The Journal of the Acoustical Society of America | AIP Publishing

I don’t know if the study is proper science or not though, or maybe there’s some technicality that I don’t understand - I’d like to hear what you think. In any case, I think you’re right that the higher frequencies are more sensations than actual sounds.

Cheers.

I’m probably not getting something, but this is the spectrogram/frequenciy analysis that I’m getting (recording through either microphone or speakers):
http://imgur.com/smFAXtS
The recording is of the 15-19 kHz part. What am I missing?

To return on-topic though. Ideally I’d like to remove all the noise and isolate the high pitched frequency (-ies) and turn up the volume - without slowing down the recording. That way the repair guys can’t claim that the sound is “constructed” in Audacity by manipulating the clip (other than removing noise). Is this possible? Could one of you maybe use one of my clips and do this, or tell me how?

Wikipedia can be misleading, and in this case I think it is misleading.
Response to frequencies well above 20 kHz has been recorded when transmitted through bone conduction. In the tests referred to by Wikipedia, the test subject had their head “attached to a headrest” and tones blasted at them at 90 dB from 50 cm distance. This clearly does not relate to the common day to day experience that we refer to as “hearing”. Going even higher in frequency and increasing the power, it would be possible to burn the subjects ears due to the thermal effects of ultrasound - sure the ultrasound would be “detected” by the subject, but would we call that “hearing”?

In the “Take 1, Wasapi.wav” sample, there is clearly a tone at about 14 kHz in the right channel:

This code in the Nyquist Prompt will notch out that tone very precisely in the right channel only:

(vector (aref *track* 0)
  (notch2 (aref *track* 1) 14275 20))

If you have a longer recording of that noise, you could remove the noise for some parts of the recording and leave it in for other parts. Then when you listen you may be able to hear the noise start and stop.

I just double-checked YouTube v=VxcbppCX6Rk at HD 1080p , nothing above 16kHz when played on YouTube on my machine. Maybe if that video is downloaded, (rather than streamed), you get the original video , (not a transcoded version), which could have a wider frequency range than what you get when playing (the transcoded version) on YouTube.

Which is why I had a look at the source material. But a layman like myself can’t really interpret all that information easily.

No, I suppose we wouldn’t. I guess I didn’t consider that there’s a line between hearing and detecting. I imagine it’s blurry though. Thanks for the explanation!

Thank you, this will go a long way! But is it possible to do the opposite? To leave only the high pitched sound?

I didn’t download the video though, I just hit “play” on Youtube and then “record” on Audacity. I guess we’re doing something differently?

Ultra Narrow filter will isolate a tone , but if you applied such a filter to plain white-noise, (without any tone in it) , the result will be a constant tone. So the result of Ultra Narrow is not good evidence tone exists.

These are the tones embedded in the YouTube video.

7980 -42.913
8000 -16.9555
8020 -42.8148

11980 -44.0055
12000 -17.3734
12020 -44.0486

14980 -72.9205
15000 -16.9237
15020 -73.2234

15980 -73.5749
16000 -12.0409
16020 -74.6918

16980 -37.9174
17000 -12.5455
17020 -37.8853

17980 -33.2372
18000 -12.1389
18020 -33.2445
(Hz and dB)

There’s also a 16 kHz tone at the end of the video.

For best visualisation (guesswork…): resample to 409600 Hz, rectangular window, size 2048 (=20 dB steps).
To check the presence of a tone: change pitch by -12 dB.

Robert

I understand, thanks.

Thanks, I’ll look into that!

---

I think I have everything that I need now, so thank you to everyone for your help (and discussions)! Great forum! I might return to bother you if there’s anything else