Test track audibility

[gfinlays]

I wasn't entirely sure of where to post this. In relation to a discussion elsewhere regading the presence or lack of audible effects of different ethernet equipment (most of the claims talk about "noise"), I recorded some test tracks and listened to them. I worked on the premise that if "noise" over ethernet was really a thing, then my entirley commodity network setup would make the -120 dBFS test tracks inaudible above the noise floor.

Here's a link to one of the test tracks I used: https://www.dropbox.com/s/tnfc57u81f9xrbi/Test tone.flac?dl=0

The test track is quite likely to be buried in the noise floor of many systems, others may not be able to apply sufficeint gain to make it audible, so I'm curious as to how many folks can actually hear it and describe it's nature. Please download, have a listen and vote in the poll.

 

[Doodski]

I am 1:20 in and hear the 200mm fans of my desktop system. It's dead silent for me at the end.

 

[staticV3]

Honestly, what's the point? To find out whose system has a ridiculous 100+dB headroom?


My answer is none of the above btw.

 

[gfinlays]

I am 1:20 in and hear the 200mm fans of my desktop system. It's dead silent for me at the end.

Once I have a good number of replies, I'll fess up and reveal the nature of the tone. I can hear it with my ear pressed against the tweeter of my speakers.

 

[gfinlays]

Honestly, what's the point? To find out whose system has a ridiculous 100+dB headroom?
View attachment 248901

The answer is none of the above btw.

The point is that some claim sonic improvements from "audiophile" network gear on the premise that lesser gear introduces noise into the playback stream reproduction chain which masks "detail". I set about testing the audible limitations of my system (all of the source componentry beyond the streamer has been tested and reviewed by @amirm) - more on that later. Given we're a more objective lot over here, with many of us having chosen kit based on objective performance rather than subjective reviews, I'm trying to gauge a rough percentage of who can hear the nature of the test tone on their system. I can hear it on mine at close to maximum gain and there isn't a shred of "audiophile" network gear in sight.

This is purely about scientific interest, audibility in relation to system noise floor and exposing the claims around "audiophile" ethernet.

The answer is one of the above btw.

 

[Doodski]

OK then. Lets get to the next phase of testing.

 

[kemmler3D]

The point is that some claim sonic improvements from "audiophile" network gear on the premise that lesser gear introduces noise into the playback stream which masks "detail". I set about testing the audible limitations of my system (all of the source componentry beyond the streamer has been tested and reviewed by @amirm) - more on that later. Given we're a more objective lot over here, with many of us having chosen kit based on objective performance rather than subjective reviews, I'm trying to gauge a rough percentage of who can hear the nature of the test tone on their system. I can hear it on mine at close to maximum gain and there isn't a shred of "audiophile" network gear in sight.

The answer is one of the above btw.

Should be audible on any reasonable system - with enough gain. Tones are discernable even with only 2-3 bits available. System noise (even if it is well above the signal) doesn't actually make most signals inaudible. You just hear a lot of noise at the same time. If your test track is actually pink or white noise, the previous sentence might not apply. It's hard to tell noise apart from noise regardless of levels.

To put it another way, you do not actually need noise to be below -120dB FS to hear a tone that's also at -120dBFS, just like noise at -12dBFS doesn't fully mask any content at -13dBFS and below. The masking threshold corresponds to Fletcher-Munson and ranges from roughly 0 - 60dB depending on frequency. https://en.wikipedia.org/wiki/Maski...ng threshold is,two sounds close in frequency.

To put it ANOTHER way, depending on what the mystery signal is, your SINAD might need to be as bad as 60dB to fully mask the mystery signal and make it inaudible.

 

[gfinlays]

Should be audible on any reasonable system - with enough gain. Tones are discernable even with only 2-3 bits available. System noise (even if it is well above the signal) doesn't actually make most signals inaudible. You just hear a lot of noise at the same time. If your test track is actually pink or white noise, the previous sentence might not apply. It's hard to tell noise apart from noise regardless of levels.

So give the track a whirl and hit me with your conclusion

 

[staticV3]

I'm trying to gauge a rough percentage of who can hear the nature of the test tone on their system.

That would be fine if only you didn't say

The test track is quite likely to be buried in the noise floor of many systems, others may not be able to apply sufficeint gain to make it audible, so I'm curious as to how many folks can actually hear it

Now it's no longer a "whose system is quiet enough to resolve the signal?", but a "who can apply enough gain to the signal to get it above their system's noise floor?"

Anyone can hear the signal if they just enable Replay Gain.

You should have specified something like "Set volume to normal levels, then play the test signal without changing volume, no replay gain."

The answer is one of the above btw.

It's not.

 

[Doodski]

All my PC volume settings are @ 100%. The Schiit Hersey headphone amp is at maybe 2pm and that only tells me things are close'ish to maxed. I can hear the PC fans clear as a bell with Sennheiser HD598 SR headphones when your test track is playing..

Sennheiser HD 598SR - Headphone Around Ear - Best Audio Quality

HD 598SR - Headphone Around Ear - Developed for home - HiFi Stereo Best Audio Quality - Sennheiser Discover True Sound - Sennheiser’s ’Ergonomic Acoustic Refinement' E.A.R. technology - sennheiser.com

en-ca.sennheiser.com

 

[gfinlays]

That would be fine if only you didn't say


Now it's no longer a "whose system is quiet enough to resolve the signal?", but a "who can apply enough gain to the signal to get it above their system's noise floor?"

Anyone can hear the signal if they just enable Replay Gain.

You should have specified something like "Set volume to normal levels, then play the test signal without changing volume, no replay gain."


It's not.

My point is that audibility isn't just about noise floor. It's also about signal gain and absolute listening level. If you can pull a -120 dBFS tone up to audible levels and discern the nature of the tone, you have a low noise floor system which is objectively "highly resolving" as the "audiophphools" [sic] like to say. Furthermore, at non-hearing damaging levels, noise is significantly below the threshold of audibility.

 

[Doodski]

My point is that audibility isn't just about noise floor. It's also about signal gain and absolute listening level. If you can pull a -120 dBFS tone up to audible levels and discern the nature of the tone, you have a low noise floor system which is objectively "highly resolving" as the "audiophphools" [sic] like to say. Furthermore, at non-hearing damging levels, noise is significantly below the threshold of audibility.

Huh?

 

[staticV3]

If you can pull a -120 dBFS tone up to audible levels and discern the nature of the tone, you have a low noise floor system which is objectively "highly resolving" as the "audiophphools" [sic] like to say.

I can hear the test signal on my phone's garbage internal speaker. Now what? Does that make it highly resolving?

 

[gfinlays]

Huh?

I should have qualified that by saying that when listening to music at non hearing damaging levels, the noise level of the system is way below the threshold of audibility.

 

[restorer-john]

Personally, I wouldn't be listening to any random 'test' tracks somebody posts on the internet. Not without at least giving it a brief analysis.

It's easy to make a trojan horse style track with an extremely low level tone and a high level 20kHz tone buried in it. You know, the track that vaporises your tweeters while you wind up the volume to listen to the low level tone? This isn't such a track, but you have been warned people...

This track appears to be this (I've normalized the track to see it):

 

[staticV3]

This track appears to be this:

which is none of the options in the poll @gfinlays

 

[Doodski]

Personally, I wouldn't be listening to any random 'test' tracks somebody posts on the internet. Not without at least giving it a brief analysis.

It's easy to make a trojan horse style track with an extremely low level tone and a high level 20kHz tone buried in it. You know, the track that vaporises your tweeters while you wind up the volume to listen to the low level tone? This isn't such a track, but you have been warned people...

This track appears to be this:

View attachment 248907

I use my gaming PC/experimental and research box. All personal matters go onto the notebook in case I need to bug out.

 

[gfinlays]

Personally, I wouldn't be listening to any random 'test' tracks somebody posts on the internet. Not without at least giving it a brief analysis.

It's easy to make a trojan horse style track with an extremely low level tone and a high level 20kHz tone buried in it. You know, the track that vaporises your tweeters while you wind up the volume to listen to the low level tone? This isn't such a track, but you have been warned people...

This track appears to be this (I've normalized the track to see it):

View attachment 248907

Well, that burst the bubble.

No, it's not some trojan horse style track intended to vapourise anyone's tweeters, it's a -120 dBFS 1 kHz square wave, so a single fundamental frequency test tone. If anyone really wants to split hairs, a square wave isn't purely a single frequency test tone - as most of us know it's made up of a fundamental plus 1/n amplitude odd order harmonics of the fundamental.

Anyway, my point was about trying to establish whether many people could 1. pull the test signal into audibility and 2. if they could, was the nature of the tone discernable?

I streamed the test tracks (I tried quite a few different ones) from my Roon local library via a PoE powered RPi4 feeding a Benchmark DAC3 HGC > HPA4 > dual mono AHB2s.

Taking this into account, the arguments around network equipment affecting sound quality (based on the usual hand-waving pseudo-scientific technobabble oft-repeated without any shred of objective evidence by audiophiles) are clearly nonsense.

If the system is capable of making a -120 dBFS signal audible, then ethernet borne noise is clearly way below the threshold of audibility at non-deafness inducing playback SPLs when listening to real music.

 

[kongwee]

Can't hear anything maxing out my iMac. On my audio interface, don't move a single meter. With a test osc -67db on DAW and -6dB on my audio interface that peak on 7500khz test tone. Actually have to increase the range to get pass the ambient noise, but I don't have sound meter on hand let say 60dB SPL home noise. It is a range of more than 120dB. Of course, this is unscientific way to calculate. Nobody I have found to translate dBv into dB SPL.

 

[kemmler3D]

If the system is capable of making a -120 dBFS signal audible, then ethernet borne noise is clearly way below the threshold of audibility at non-deafness inducing playback SPLs when listening to real music.

Although I agree with your overall conclusion - ethernet borne noise is not important - one does not follow from the other. Just because we can apply enough gain to a -120dB sound to make it audible above noise, doesn't imply you can or can't hear an unknown source of noise on the same system.