Beta Test: DISTORT - audibility of distortions

[Blumlein 88]

Well, that tells me I do need a blind comparator in DISTORT

FFTs are used for display only and to compute THD and harmonic content. They are not used for playback or to distort a file. At all

That should have been obvious with a few seconds thought.

Let me see.........
it was late,
I'd drunk too much (Pepsi in this case, my hearing was all jittery from the caffiene),
I was listening over speakers and not headphones,
my good USB cable was in the shop getting a tune up,
the wind was from the north instead of the normal south-west causing strange resonances in the house,
I'm sure I'm forgetting other important contingencies........

 

[Blumlein 88]

OK, so something like a sone scale instead of dB?

I like the idea of a Sone scale.

 

[Newk Yuler]

This program has commercial value. Especially if simplified for use by regular folks. Such as an easy to apply realtime filter for audio or a complete player with presets.

Would it be possible and beneficial to implement Nvidia CUDA processing (such as used in HQPlayer) since the program is so processor intensive?

 

[RayDunzl]

Can you add a mode that shows the decibel axis as logarithmic?

"The decibel (dB) is a logarithmic unit used to measure sound level."

You wanna double that up?

 

[Theriverlethe]

"The decibel (dB) is a logarithmic unit used to measure sound level."

You wanna double that up?

Would that make it worse?

 

[Theriverlethe]

OK, so something like a sone scale instead of dB?

Aren’t sones derived from equal-loudness contours? This could get complicated quickly.
I just meant something like an impulse at -10dB would be half the height of 0dB, -20dB would be half the height of -10dB, and so on. This might give neophytes a more accurate conceptual framework.

 

[RayDunzl]

Would that make it worse?

I dunno.

Show a plot with decibeled decibels...

 

[Theriverlethe]

I dunno.

Show a plot with decibeled decibels...

Jeez, man, I just had to learn what a logarithm is. Can’t we eat mushrooms and listen to different film capacitors instead?

 

[pkane]

This program has commercial value. Especially if simplified for use by regular folks. Such as an easy to apply realtime filter for audio or a complete player with presets.

Would it be possible and beneficial to implement Nvidia CUDA processing (such as used in HQPlayer) since the program is so processor intensive?

Well... if I were to make it into a commercial product, sure CUDA, maybe even quantum computing would be considered

 

[pkane]

I like the idea of a Sone scale.

@Theriverlethe and @Blumlein 88 , you made me curious, so here's Sone scale version. I guess it more clearly indicates that some of the harmonics are really much lower in perceptual volume than the normal dB scale would indicate. But then, there are many things this doesn't take into account, like different frequencies of the harmonics (Sone scale is meant for 1kHz frequency). I also assumed that the fundamental harmonic is playing at 90dB level (=0dBFS) so that the perceptual scale could be applied.

Here's the (original) dB version:

And here's the same version, except plotted on a Sone scale (I truncated all content below 0dB):

If I'm reading this correctly, this indicates that the fundamental is much more perceptually obvious (loud) than even the fairly high second harmonic that's at -16dBFS. Interesting!

 

[Theriverlethe]

@Theriverlethe and @Blumlein 88 , you made me curious, so here's Sone scale version. I guess it more clearly indicates that some of the harmonics are really much lower in perceptual volume than the normal dB scale would indicate. But then, there are many things this doesn't take into account, like different frequencies of the harmonics (Sone scale is meant for 1kHz frequency). I also assumed that the fundamental harmonic is playing at 90dB level (=0dBFS) so that the perceptual scale could be applied.

Here's the (original) dB version:
View attachment 41329

And here's the same version, except plotted on a Sone scale (I truncated all content below 0dB):
View attachment 41328

If I'm reading this correctly, this indicates that the fundamental is much more perceptually obvious (loud) than even the fairly high second harmonic that's at -16dBFS. Interesting!

That looks like what I was thinking. Thanks.

 

[Blumlein 88]

Well you really need to map the harmonics on a phon scale and convert to sones for it to be correct. The phons uses Fletcher-Munson equal loudness contours and then those are used to map sone scaling. Are you doing that in your mapping of the harmonics?
@pkane

 

[pkane]

Well you really need to map the harmonics on a phon scale and convert to sones for it to be correct. The phons uses Fletcher-Munson equal loudness contours and then those are used to map sone scaling. Are you doing that in your mapping of the harmonics?
@pkane

As I said, the conversion only works correctly for a 1kHz signal. All others will be off by a bit. I don't think it's worth it for me to put in the full spectrum equal loudness curves, not sure if it'll reveal anything more useful. Maybe when I have nothing better to do

 

[Blumlein 88]

As I said, the conversion only works correctly for a 1kHz signal. All others will be off by a bit. I don't think it's worth it for me to put in the full spectrum equal loudness curves, not sure if it'll reveal anything more useful. Maybe when I have nothing better to do

Yes that would be quite a bother to do. And I'm not trying to come up with extra work for you. Was just curious which way you were doing this.

All we need in the end is another bit of software for you to develop. Take a piece of music, analyze the spectrum, apply a phon scale to it, implement masking curves, and tell us with a given distortion profile imported from Distort if the music is reproduced transparently or not for a young human listener with undamaged hearing. That should be a piece of cake as long as you develop it.

 

[Theriverlethe]

Yes that would be quite a bother to do. And I'm not trying to come up with extra work for you. Was just curious which way you were doing this.

All we need in the end is another bit of software for you to develop. Take a piece of music, analyze the spectrum, apply a phon scale to it, implement masking curves, and tell us with a given distortion profile imported from Distort if the music is reproduced transparently or not for a young human listener with undamaged hearing. That should be a piece of cake as long as you develop it.

Equal loudness contours vary with SPL, so you’d also have to know what SPL a 0dBFS signal creates.

 

[Blumlein 88]

Equal loudness contours vary with SPL, so you’d also have to know what SPL a 0dBFS signal creates.

Simple adjustment in a drop down menu. You could choose the value that is the max output from your speaker.

There are of course possible additional enhancements. The software could also use measured FR of your speakers to make even finer adjustments to determine if you'll transparently hear the analyzed bit of music and a given distortion profile.

 

[Arpiben]

@pkane

Hi Paul,
Thanks to Distort and @Blumlein 88 I finally joined ASR.

Dealing with v1.0.10 it seems that:

• Help/About is linking to the wrong page
• Mains @50/60 Hz are not correctly represented in the spectrum view but properly computed with the file

Adding some modulated noise floor could be interesting too in terms of judging its audibility.
Once again, Great job Paul. Thanks.

 

[pkane]

@pkane

Hi Paul,
Thanks to Distort and @Blumlein 88 I finally joined ASR.

Dealing with v1.0.10 it seems that:

• Help/About is linking to the wrong page
• Mains @50/60 Hz are not correctly represented in the spectrum view but properly computed with the file

Adding some modulated noise floor could be interesting too in terms of judging its audibility.
Once again, Great job Paul. Thanks.


View attachment 41517

View attachment 41518

Hi Arpiben, nice to see you here!

Wow, how did I miss this frequency error? I swear I tested noise settings and saw it working! Must be expectation bias. I'll fix this asap.

What would you use to modulate the noise floor? The signal itself, or just arbitrary frequency sine wave, or maybe a completely independent recorded file?

I'm working on the next distortion that you'll appreciate: jitter. Looking forward to your feedback on that one! Here's a little preview:

 

[Blumlein 88]

@pkane

Hi Paul,
Thanks to Distort and @Blumlein 88 I finally joined ASR.

Dealing with v1.0.10 it seems that:

• Help/About is linking to the wrong page
• Mains @50/60 Hz are not correctly represented in the spectrum view but properly computed with the file

Adding some modulated noise floor could be interesting too in terms of judging its audibility.
Once again, Great job Paul. Thanks.


View attachment 41517

View attachment 41518

Glad you are here, and will be taking part in ASR forums.

 

[Arpiben]

Hi Arpiben, nice to see you here!

What would you use to modulate the noise floor? The signal itself, or just arbitrary frequency sine wave, or maybe a completely independent recorded file?

Good questions Paul ?
In fact I need to do some research before. "Noise Floor Modulation" verbiage of some DAC designers is puzzling me.
Eventually, @March Audio may help us since he started a thread a few years back here.
Distort software can turn a good tool for easily checking such claims