dBV is actually mislabeled dBFS by Umcaruje · Pull Request #3095 · LMMS/lmms

Umcaruje · 2016-10-28T22:50:30Z

So, after some looking around, I realised LMMS already used dBFS values but labeled them as dBV. according to wikipedia, dBV is a decibel that's relative to 1V of voltage, which does not quite fit in software.

dBFS on the other hand is relative to the maximum value of a digital full scale, which in our case is 1 in the linear scale, so the conclusion is that we are indeed actually using dBFS in LMMS.

Also @michaelgregorius also came to this conclusion in #2672 but probably never got around to this rename job.

BaraMGB · 2016-10-28T23:06:38Z

Agreed!

tresf · 2016-10-28T23:21:00Z

+	qa->setData("displaydbfs");
 	qa->setCheckable( true );
-	qa->setChecked( ConfigManager::inst()->value( "app", "displaydbv" ).toInt() );
+	qa->setChecked( ConfigManager::inst()->value( "app", "displaydbfs" ).toInt() );


Is this going to need an upgrade()?

Do we even do upgrades for .lmmsrc.xml? If so, yes I can write an upgrade.

Albeit rather recently, yes we do.

Ok, I'll look into it.

@tresf done per 5a7acd3

tresf · 2016-11-01T02:12:22Z

I don't know enough about dbfs to merge, but otherwise the PR is good to merge. I did notice some inconsistent whitespace formatting that this commit introduces. We should probably decide eventually which one of these is correct, but this PR isn't the place for that decision.

     if (foo) {
           // foo
+          whitepsace option 1
+whitespace option 2
           // bar
     }

BaraMGB · 2016-11-01T09:07:33Z

For what I've learned in Wikipedia dB FS is the unit used by all digital audio signals. For analog signals is dbv used (voltage).

Umcaruje · 2016-11-03T21:35:27Z

@tresf please comment on the code where you see that issue. This is a simple find/replace job, and I didn't touch the whitespace.

tresf · 2016-11-03T22:37:31Z

 		return;
 	}
-
+	


Oh, ok, I accidentaly added whitespace there. I'll remove it.

zonkmachine · 2017-03-08T01:28:56Z

 	painter.drawPixmap( drawRectR, *m_leds, drawRectR );

-	float const persistentRightPeakDBFS = ampToDbv(m_persistentPeak_R);
+	float const persistentRightPeakDBFS = ampToDbfs(m_persistentPeak_R);


ampToDbfs( qMax<float>( 0.0001, m_persistentPeak_R ) ) ?
@Umcaruje ampToDbfs( Must_be_greater_than_zero! )
I'm debugging with a new secret weapon and you've got a floating point error here. I triggered it by just starting lmms in gdb with @michaelgregorius super hack and when it loaded I just launched the Empty template. Boom! Stuck the qMax code in there and the issue went away. Care to take a look at this?

Program received signal SIGFPE, Arithmetic exception. 0x00007ffff47d51ac in feraiseexcept (__excepts=4) at ../sysdeps/x86/fpu/bits/fenv.h:135 135 ../sysdeps/x86/fpu/bits/fenv.h: No such file or directory. (gdb) bt #0 0x00007ffff47d51ac in feraiseexcept (__excepts=4) at ../sysdeps/x86/fpu/bits/fenv.h:135 #1 __log10f (x=0) at w_log10f.c:32 #2 0x00000000005ec737 in ampToDbfs (amp=0) at /home/zonkmachine/builds/lmms/lmms/include/lmms_math.h:273 #3 0x00000000005ee1eb in Fader::paintDBFSLevels (this=0x1650ef0, ev=0x7fffffffb5f0, painter=...) at /home/zonkmachine/builds/lmms/lmms/src/gui/widgets/Fader.cpp:409

We should use safeAmpToDbfs here then? That has built in protection for zeores. I'll take a more thorough look soon.

We should use safeAmpToDbfs here then?

Yes. This looks a bit odd however: https://github.com/LMMS/lmms/blob/master/include/lmms_math.h#L251
amp == 0.0f There is no test here for negative input. I'm not up to date on the math here but this reference hints at an error for negative arguments.
http://www.cplusplus.com/reference/cmath/log10/

If x is negative, it causes a domain error.

Yes, but we never use negative values for linear amplitude though, so it shouldn't really be even considered. scratches head and the variables are initialized with the value of zero: https://github.com/Umcaruje/lmms/blob/e3aad60661b9581f12e3d91bfc2158618e1ca571/src/gui/widgets/Fader.cpp#L73-L74

Though you loaded the Empty template, and there are no instruments there. Does this happen with the default template?

I'm failing to reproduce that right now as I just get the crash right away. I think I might have mistaken and just thought I managed to launch something while it had already closed down business. Anyway.
Line 387 reads: float const persistentLeftPeakDBFS = ampToDbfs(m_persistentPeak_L);
The bt full readings for line 387 reads:

#3 0x00000000005ee070 in Fader::paintDBFSLevels (this=0x16556b0, ev=0x7fffffffbb90, painter=...) at /home/zonkmachine/builds/lmms/lmms/src/gui/widgets/Fader.cpp:387 peak_L = 0 persistentLeftPeakDBFS = -nan(0x7fbba4) width = 11 center = 12 maxDB = 9 leftSpan = 0 rightSpan = 4.59163468e-41 drawRectR = {x1 = 0, y1 = 0, x2 = 0, y2 = 0} persistentPeak_R = 32767 peak_R = -18464 persistentRightPeakDBFS = 4.59163468e-41 height = 116 minDB = -42.0000038 fullSpanReciprocal = 0.019607842 drawRectL = {x1 = 0, y1 = 116, x2 = 10, y2 = 115} persistentPeak_L = -17520

int persistentPeak_L = height * (1 - (persistentLeftPeakDBFS - minDB) ...

It's overflowing.
~~unsigned int persistentPeak_L seem to be a fix. Not tested this a lot though.~~

If I start up and press play directly it's cool. If I wait a couple of seconds and do nothing it crashes.

Shouldn't we rather open a new issue for this one?

Yes, you're right.

Continued here: #3420

unfa · 2017-03-08T10:59:24Z

I think that in the FX Mixer, it'd make more sense to label the fader positions using "dB" units not "dBFS", because then +6 dBFS sounds weird. +6dB is perfectly fine though.

* Relabel "dBV" to "dBFS" in function names and GUI * Write a ConfigManager upgrade for older versions

dBV is actually mislabeled dBFS

ea6921a

tresf reviewed Oct 28, 2016

View reviewed changes

Write a ConfigManager upgrade

5a7acd3

tresf reviewed Nov 3, 2016

View reviewed changes

Fix whitespace

e3aad60

tresf merged commit c961896 into LMMS:master Nov 7, 2016

Umcaruje deleted the dbvtodbfs branch November 10, 2016 21:41

zonkmachine reviewed Mar 8, 2017

View reviewed changes

zonkmachine mentioned this pull request Mar 13, 2017

Fader.cpp - Floating point error #3420

Closed

sdasda7777 pushed a commit to sdasda7777/lmms that referenced this pull request Jun 28, 2022

dBV is actually mislabeled dBFS (LMMS#3095)

7ccab22

* Relabel "dBV" to "dBFS" in function names and GUI * Write a ConfigManager upgrade for older versions

Uh oh!

Conversation

Umcaruje commented Oct 28, 2016

Uh oh!

BaraMGB commented Oct 28, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Umcaruje Oct 31, 2016 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

tresf commented Nov 1, 2016 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

BaraMGB commented Nov 1, 2016

Uh oh!

Umcaruje commented Nov 3, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

zonkmachine Mar 8, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

unfa commented Mar 8, 2017

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

Umcaruje Oct 31, 2016 •

edited

Loading

tresf commented Nov 1, 2016 •

edited

Loading

zonkmachine Mar 8, 2017 •

edited

Loading