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: Strange frequency spike near 18kHz in song (Read 8010 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Strange frequency spike near 18kHz in song

Hello, all.

I apologize if this isn't the appropriate subforum, I couldn't find a more suitable one.

As I was listening to Sonic Mania's FLAC soundtrack (provided via a download code after one purchases the vinyl), I noticed something odd in foobar's spectrogram when one of the songs was playing. Track 9, Comfort Zone (Main Menu), has a strange frequency that's more prominent than it should be.

So I used SoX and Audacity to generate a spectrogram and a frequency plot, respectively, and this is what I found.





There seems to be a peak near 18kHz.

Anyone have any idea what could have caused this?

Thanks.
Lossless: flac --best --verify
Lossy: opusenc --bitrate 160

Re: Strange frequency spike near 18kHz in song

Reply #1
Quote
Sonic Mania's FLAC soundtrack
I don't know what that is, but it's "artificial music" (digital/electronic/synthesizer) right?

Does it sound OK?

It's not continuous so it's probably not noise or interference.     I'd guess it's a (artificial) high-hat or some other intentional percussive sound on the beat. 

Re: Strange frequency spike near 18kHz in song

Reply #2
Quote
Sonic Mania's FLAC soundtrack
I don't know what that is, but it's "artificial music" (digital/electronic/synthesizer) right?

That would seem to be the case for most of the album, yes. There's some songs here and there where a real electric guitar is seemingly used, but not this track; most of the album seems to use synths/VSTs. I would upload a 30-second sample but the wording of TOS9 makes me wonder if I'm only allowed to do so for the purpose of codec testing, and not for the purposes of this specific thread (which isn't codec testing, but so that you can see if it's "artificial music" or not).

Does it sound OK?

It's not continuous so it's probably not noise or interference.     I'd guess it's a (artificial) high-hat or some other intentional percussive sound on the beat. 

It does indeed sound OK, like, there's doesn't seem to be anything unnatural or distorted; it all sounds musical. And you're right about it not being continuous, most notably at the end when the track ends and the synths begin to fade out. This is coincidentally also when the percussion stops playing, so it being a hi-hat makes sense — but what hi-hat is concentrated close to 18kHz, even if artificial? I mean, most people over the age of 30 can't even hear that.

I guess if I can upload 30-second sample, it may help diagnose what exactly this thing is. If I can, let me know which segment of the track would prove most useful: the ending 30 seconds, or the first 30 seconds.
Lossless: flac --best --verify
Lossy: opusenc --bitrate 160

Re: Strange frequency spike near 18kHz in song

Reply #3
I'm familiar with synthesizers, it could been how someone decided to bend a certain sample, either by the person that produced the soundfont or synthesizer used in the song or by someone tweaking the parameters of the said synthesizer in question and decided they liked how it sounded?  Producing music with synthesizers is different than recording instruments through a microphone.

Re: Strange frequency spike near 18kHz in song

Reply #4
I'm familiar with synthesizers, it could been how someone decided to bend a certain sample, either by the person that produced the soundfont or synthesizer used in the song or by someone tweaking the parameters of the said synthesizer in question and decided they liked how it sounded?  Producing music with synthesizers is different than recording instruments through a microphone.

That's certainly plausible, but again, it's how high this frequency is what seemed odd to me. It's in the range of frequencies where a middle-aged person probably wouldn't even be able to hear it, so it seems odd to me that a musical choice would be made involving frequencies that are barely audible or not audible at all to a significant part of the population.
Lossless: flac --best --verify
Lossy: opusenc --bitrate 160

Re: Strange frequency spike near 18kHz in song

Reply #5
That's certainly plausible, but again, it's how high this frequency is what seemed odd to me. It's in the range of frequencies where a middle-aged person probably wouldn't even be able to hear it, so it seems odd to me that a musical choice would be made involving frequencies that are barely audible or not audible at all to a significant part of the population.

One of my neighbors couldn't understand a high pitched vocalist on the radio but her daughter could and this singer was in her early twenties and the daughter was in her preteens.  The vocalist is one of the biggest names in the type of music she produces and has even been in a few T-Mobile commercials, the last couple of years.

The vast majority of people just listen to their music to enjoy it and don't even think about looking at the spectrogram.  Because someone is unable to hear certain high-pitched sounds because of their age doesn't mean that can't enjoy it.  If the music is produced by someone that is young, then it's likely that they weren't even aware or cared if some middle age person could hear all of it or not.

I tend to think of music as one's (or a group's) auditory art and crafts.


Re: Strange frequency spike near 18kHz in song

Reply #7
I'm going with a (benign) bug in the percussion synth.

Conceivably, if the synth is based on a sample, then it could be an interfering tone captured with the sample, but this seems less likely to me.

Re: Strange frequency spike near 18kHz in song

Reply #8
That's certainly plausible, but again, it's how high this frequency is what seemed odd to me. It's in the range of frequencies where a middle-aged person probably wouldn't even be able to hear it, so it seems odd to me that a musical choice would be made involving frequencies that are barely audible or not audible at all to a significant part of the population.

One of my neighbors couldn't understand a high pitched vocalist on the radio but her daughter could and this singer was in her early twenties and the daughter was in her preteens.  The vocalist is one of the biggest names in the type of music she produces and has even been in a few T-Mobile commercials, the last couple of years.

The vast majority of people just listen to their music to enjoy it and don't even think about looking at the spectrogram.  Because someone is unable to hear certain high-pitched sounds because of their age doesn't mean that can't enjoy it.  If the music is produced by someone that is young, then it's likely that they weren't even aware or cared if some middle age person could hear all of it or not.

I tend to think of music as one's (or a group's) auditory art and crafts.

All of this in principle is true (as I said, the music sounds fine), but again, look at the spike in the Audacity frequency analysis and how high it is in frequency. Using your example of a high-pitched vocalist — to put this into perspective, a standard 88-key piano's highest note is the C8, corresponding to a (fundamental) frequency of 4.186kHz. The vocalist your neighbor couldn't hear certainly sang lower than this, and our frequency (the spike in the song) is closer to 18kHz!

Possibly a clue:

https://hydrogenaud.io/index.php/topic,17212.0.html

I've thought about this, but aren't those sorts of tones a continuous line throughout the spectrum, and at a somewhat lower frequency? Also, as was stated before, since this album is mostly "artificial" (made with synths/VSTs), I'm not sure where in the mixing/mastering/production chain such a spurious interference could have found its way onto the track.

I'm going with a (benign) bug in the percussion synth.

Conceivably, if the synth is based on a sample, then it could be an interfering tone captured with the sample, but this seems less likely to me.

Hm, now that's quite possible. As far as "benign" goes, well, for me I guess it is (I am close to middle-aged, and according to a test I did a few months ago, I can't hear past about 17.2kHz, so this tone for me would be inaudible), but as it happens with those flyback transformer on the back of TVs that are accidentally captured, for some people it can be audible and even annoying, so assuming this wasn't an intentional artistic choice and indeed some bug/glitch in the sample/VST that was used, I wonder how many people noticed and found it unmusical.

EDIT: I reiterate, if a 30-second sample can be legally uploaded (in accordance with TOS9) that could help elucidate this situation, by all means, let me know.
Lossless: flac --best --verify
Lossy: opusenc --bitrate 160

Re: Strange frequency spike near 18kHz in song

Reply #9
It seems not uncommon to have high level inaudible frequencies in electronic music. Two more examples.
Spoiler (click to show/hide)

 

Re: Strange frequency spike near 18kHz in song

Reply #10
It seems not uncommon to have high level inaudible frequencies in electronic music. Two more examples.
Spoiler (click to show/hide)

Those are indeed very interesting examples.

I'll note one thing: unlike my track, the high-frequency content in these two songs seems to be continuous and not intermittent, so it doesn't suggest a percussion synth (buggy or intentional).

Would it be possible for you to do a frequency analysis like I did in Audacity for these two tracks of yours? It may be useful to see how the frequency is distributed; in most other analyses I've performed, the audio seems to be distributed roughly evenly with a slight bias towards the lower frequencies; this is one of the reasons the very prominent spike near 18kHz surprised me, though I admit I haven't performed many frequency analyses for electronic/synth-based music.
Lossless: flac --best --verify
Lossy: opusenc --bitrate 160

Re: Strange frequency spike near 18kHz in song

Reply #11
Would it be possible for you to do a frequency analysis like I did in Audacity for these two tracks of yours?
Since Audacity is unable to analyze whole tracks, because they are long, only parts with loud high frequencies are analyzed:
Spoiler (click to show/hide)

Re: Strange frequency spike near 18kHz in song

Reply #12
Would it be possible for you to do a frequency analysis like I did in Audacity for these two tracks of yours?
Since Audacity is unable to analyze whole tracks, because they are long, only parts with loud high frequencies are analyzed:
Spoiler (click to show/hide)

Hm, very interesting. So the higher frequencies get progressively more prominent all the way up the Nyquist frequency. In my track, the spurious frequency manifests as a peak, with lower values in its immediate vicinity; note the downward trend from the peak all the way to the Nyquist frequency.

I think from this we can probably conclude that these are two different phenomena, right? One (mine) is potentially a spurious frequency introduced due to whatever reason, the other (Rollin's two tracks) seem to have had a liberal and deliberate amount of EQ boost applied in the treble frequencies due to artistic choice/"sounded better to the artist and/or the engineer".
Lossless: flac --best --verify
Lossy: opusenc --bitrate 160

Re: Strange frequency spike near 18kHz in song

Reply #13
Looks like it's been amplified.


Re: Strange frequency spike near 18kHz in song

Reply #15
... the other (Rollin's two tracks) seem to have had a liberal and deliberate amount of EQ boost applied in the treble frequencies due to artistic choice/"sounded better to the artist and/or the engineer".

That is noise shaped dither.
Regards,
   Don Hills
"People hear what they see." - Doris Day

Re: Strange frequency spike near 18kHz in song

Reply #16
... the other (Rollin's two tracks) seem to have had a liberal and deliberate amount of EQ boost applied in the treble frequencies due to artistic choice/"sounded better to the artist and/or the engineer".

That is noise shaped dither.

Some of the frequency analysis could have contained in it noise-shaped dither, but if you look at the actual spectrograms, there's segments of the songs that seem to have a deliberately loud section in the higher frequencies; the fuzzy, purple mist does seem to be noise-shaped dither, but I mean the red or bright sections.
Lossless: flac --best --verify
Lossy: opusenc --bitrate 160

Re: Strange frequency spike near 18kHz in song

Reply #17
That is noise shaped dither.
There is noise-shaped dither, but high frequencies have such high level not because of dither. Shaped dither noise is ~ -90 dB.

Re: Strange frequency spike near 18kHz in song

Reply #18
Could also be an intraterestrial signal, probably from Vega.
PANIC: CPU 1: Cache Error (unrecoverable - dcache data) Eframe = 0x90000000208cf3b8
NOTICE - cpu 0 didn't dump TLB, may be hung

Re: Strange frequency spike near 18kHz in song

Reply #19
If that's in hihat, it may be sample-based, and the sample could be recorded in times when CRT-monitors were in use and they often had ~16kHz vertical refresh rate affecting the recordings. That sample if transposed up a lil would also transpose up the recorded CRT interference to ~18kHz.
Or if that hihat is procedural/synth-based, the algorithms may happen to generate top content at that frequency.

Re: Strange frequency spike near 18kHz in song

Reply #20
without 3d image (which would have also time dimension; so called spectrogram) it's hard to tell.
but tonal noises starting from ~15 kHz and above are a pretty common kind of recording defects.
this is somehow a thing even now, when CRT are almost extinct. perhaps it's enough to have a CRT nearby even if not directly visible
a fan of AutoEq + Meier Crossfeed

Re: Strange frequency spike near 18kHz in song

Reply #21
Could also be an intraterestrial signal, probably from Vega.

???

If that's in hihat, it may be sample-based, and the sample could be recorded in times when CRT-monitors were in use and they often had ~16kHz vertical refresh rate affecting the recordings. That sample if transposed up a lil would also transpose up the recorded CRT interference to ~18kHz.
Or if that hihat is procedural/synth-based, the algorithms may happen to generate top content at that frequency.

Hm, that's an interesting analysis. I suppose it could be plausible.

without 3d image (which would have also time dimension; so called spectrogram) it's hard to tell.
but tonal noises starting from ~15 kHz and above are a pretty common kind of recording defects.
this is somehow a thing even now, when CRT are almost extinct. perhaps it's enough to have a CRT nearby even if not directly visible

Hm, I hadn't thought of doing a 3D spectrogram; I wonder if I can do something with GNU Octave.
Lossless: flac --best --verify
Lossy: opusenc --bitrate 160

Re: Strange frequency spike near 18kHz in song

Reply #22
It is 3d - time axis, frequency axis, and colour for signal strength.

Re: Strange frequency spike near 18kHz in song

Reply #23
Well, technically the waterfall / voiceprint is a 3D graph: time (x), frequency (y), and intensity (color).

You can do a surfc() in octave, which will give you a nice surface plot with contour lines.

Re: Strange frequency spike near 18kHz in song

Reply #24
It is 3d - time axis, frequency axis, and colour for signal strength.

Right.

Well, technically the waterfall / voiceprint is a 3D graph: time (x), frequency (y), and intensity (color).

You can do a surfc() in octave, which will give you a nice surface plot with contour lines.

All right, I'll see if I can get to that in a while; I'm going to get busy with other things this week, so it may take a while.
Lossless: flac --best --verify
Lossy: opusenc --bitrate 160