1270280 - FireFox positioned sound bug

Status: RESOLVED DUPLICATE of bug 1012610

(Core :: Web Audio, defect, P1)

Description

[Ignas Dauksa]

User Agent: Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/50.0.2661.94 Safari/537.36

Steps to reproduce:

Launch FireFox and play the positioned audio in one of my games here: www.crystalgames.space/under


Actual results:

Heard intermittent sound which kept cutting out


Expected results:

Audio plays clearly and smoothly as it does on Chrome

Comment 1

[Loic]

WFM on Win 7 with Firefox 46

Comment 2

[Ignas Dauksa]

UPDATE: try to crash the monster and the positioned sound disappears after a while. And the positioned sound works worse than Chrome in overall details(please compare the same game to see).

More info here: https://www.scirra.com/forum/viewtopic.php?t=171907 . You can download the Capx file there and open with Construct2. It comes with instructions on how to reproduce and more info.

Comment 3

[YF (Yang)]

WFM in Fx47b2 Win10.

Comment 4

[Alice0775 White]

I can reproduce the problem on Nightly49.0a1 as well as 38.8.0ESR.
Seems 24.0 ok, 25.0 bad

STR
1. Open the URL
2. Click "black hole" under the "?" mark
3. Click anywhere screen to popping up "brown rectangle"(I do not know exactly name)
4. Move "brown rectangle" to "two red rectangle"(also I do not know exactly name)
5. Repeat step.3,4,5

(optionally)
6. Click anywhere screen to popping up "brown rectangle"
7. Move "brown rectangle" to "pink hexagon"
8. Move "brown rectangle" to "two red rectangle"
9. Repeat step.3,4,5 or 6,7,8,9

Actual Results:
Sound seems to be saturated. And start stuttering.

At least, I found a old regression range. (There may be more new regression range)
Regression window:
https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=9cdd168a8518&tochange=e7fef018f421

Comment 5

[Paul Pasca[Away. Please needinfo? Paul Oiegas [:pauloiegasSV]]]

I've tested this issue on the latest FF release(46.0.1) and on the latest Nightly(49.0a1) and I was able to reproduce it. I've also performed a regression on this issue, but I'm not sure that this is the correct regression. On some builds the sound behaves differently from build to build. Ignas can you please provide more information about this issue? The sound effect that is getting interrupted is getting louder if the player is close to the monster? In some builds that sound plays at the start of the game and it is not getting interrupted.

Last good revision: 87c1796bc46c (2013-08-12)
First bad revision: d8a62355ea26 (2013-08-13)

Pushlog:
http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=87c1796bc46c&tochange=d8a62355ea26

Comment 6

[Ignas Dauksa]

(In reply to Paul Pasca[:PoollyMcklayn] from comment #5)
> I've tested this issue on the latest FF release(46.0.1) and on the latest
> Nightly(49.0a1) and I was able to reproduce it. I've also performed a
> regression on this issue, but I'm not sure that this is the correct
> regression. On some builds the sound behaves differently from build to
> build. Ignas can you please provide more information about this issue? The
> sound effect that is getting interrupted is getting louder if the player is
> close to the monster? In some builds that sound plays at the start of the
> game and it is not getting interrupted.
> 
> Last good revision: 87c1796bc46c (2013-08-12)
> First bad revision: d8a62355ea26 (2013-08-13)
> 
> Pushlog:
> http://hg.mozilla.org/mozilla-central/
> pushloghtml?fromchange=87c1796bc46c&tochange=d8a62355ea26

Please see https://www.scirra.com/forum/viewtopic.php?t=171907 . You will need Construct2 to open the project. 

The project's Capx file has all the info needed to reproduce and understand the bug

Thanks!

Comment 7

[Loic]

Alice and Paul have the same regression range, so it's surely correct and due to bug 861693.
Of course, a reduced testcase would be better than the whole video game.

Comment 8

[Ignas Dauksa]

(In reply to Loic from comment #7)
> Alice and Paul have the same regression range, so it's surely correct and
> due to bug 861693.
> Of course, a reduced testcase would be better than the whole video game.

Here is the reduced testcase: https://www.scirra.com/forum/viewtopic.php?t=171907

Comment 9

[Chris Pearce [:cpearce (Not reading bugmail)]]

(In reply to Paul Pasca[:PoollyMcklayn] from comment #5)
> I've tested this issue on the latest FF release(46.0.1) and on the latest
> Nightly(49.0a1) and I was able to reproduce it. I've also performed a
> regression on this issue, but I'm not sure that this is the correct
> regression. On some builds the sound behaves differently from build to
> build. Ignas can you please provide more information about this issue? The
> sound effect that is getting interrupted is getting louder if the player is
> close to the monster? In some builds that sound plays at the start of the
> game and it is not getting interrupted.
> 
> Last good revision: 87c1796bc46c (2013-08-12)
> First bad revision: d8a62355ea26 (2013-08-13)
> 
> Pushlog:
> http://hg.mozilla.org/mozilla-central/
> pushloghtml?fromchange=87c1796bc46c&tochange=d8a62355ea26

Note: this regression range contains the changes that enabled DirectShow MP3 playback. But we don't use DirectShow for MP3 playback on Windows 10, we use Windows Media Foundation. So I expect the same issue affects both WMF and DirectShow.

jya: Do you have an idea what's going on here? Should this have been fixed by your recent downmixing/resampling changes?

Comment 10

[Jean-Yves Avenard [:jya]]

(In reply to Ignas Dauksa from comment #8)
> (In reply to Loic from comment #7)
> > Alice and Paul have the same regression range, so it's surely correct and
> > due to bug 861693.
> > Of course, a reduced testcase would be better than the whole video game.
> 
> Here is the reduced testcase:
> https://www.scirra.com/forum/viewtopic.php?t=171907

Hi.

How do you use this?

The only think I see there is a link to https://dl.dropboxusercontent.com/u/71417957/FFSoundBug.capx

I'm not familiar with those files..

Comment 11

[Ignas Dauksa]

(In reply to Jean-Yves Avenard [:jya] from comment #10)
> (In reply to Ignas Dauksa from comment #8)
> > (In reply to Loic from comment #7)
> > > Alice and Paul have the same regression range, so it's surely correct and
> > > due to bug 861693.
> > > Of course, a reduced testcase would be better than the whole video game.
> > 
> > Here is the reduced testcase:
> > https://www.scirra.com/forum/viewtopic.php?t=171907
> 
> Hi.
> 
> How do you use this?
> 
> The only think I see there is a link to
> https://dl.dropboxusercontent.com/u/71417957/FFSoundBug.capx
> 
> I'm not familiar with those files..

Hey!

You need to download Construct2 first here https://www.scirra.com/construct2/releases ;)

Comment 12

[Jean-Yves Avenard [:jya]]

Thanks for the very fast answer... but I'm still ensure on how I can use this and the link with HTML5 / Firefox.

This is a binary program working only on Windows.

Can you provide a minimal web page that would play this file and exhibit the problem?

Comment 13

[Ignas Dauksa]

(In reply to Jean-Yves Avenard [:jya] from comment #12)
> Thanks for the very fast answer... but I'm still ensure on how I can use
> this and the link with HTML5 / Firefox.
> 
> This is a binary program working only on Windows.
> 
> Can you provide a minimal web page that would play this file and exhibit the
> problem?

Here You go ;) http://crystalgames.space/ffbug/

Comment 14

[Jean-Yves Avenard [:jya]]

Thank you

It plays fine here on mac, just at deafening levels...

Comment 15

[Jean-Yves Avenard [:jya]]

Karl, could you have a look? this is all webaudio stuff

Comment 16

[Loic]

(In reply to Ignas Dauksa from comment #13)
> (In reply to Jean-Yves Avenard [:jya] from comment #12)
> > Thanks for the very fast answer... but I'm still ensure on how I can use
> > this and the link with HTML5 / Firefox.
> > 
> > This is a binary program working only on Windows.
> > 
> > Can you provide a minimal web page that would play this file and exhibit the
> > problem?
> 
> Here You go ;) http://crystalgames.space/ffbug/

It's clearly reproducible on Win 7: FF21, no issue, FF25, bug.

Comment 17

[Alice0775 White]

Yes, I got a same regression range(comment#4) with testcase comment#13.

Reproducible: always(very easy)

Steps:
1. Open http://crystalgames.space/ffbug/
2. Repeatedly click [blue], [yellow] button alternately

Actual Results:
stuttering

Comment 18

[Karl Tomlinson (:karlt)]

I would guess this is the same issue as bug 1012610.

I think Gecko still implements the distance models more or less as spec'd here
https://www.w3.org/TR/2013/WD-webaudio-20131010/#dfn-distanceModel

When the default "inverse" model and unit rolloffFactor is used, the gain is
inversely proportional to distance, and so, when the distance is small, the
sound distorts.

Perhaps the regression range in comment 4 somehow relates to triggering
whether c2runtime.js uses WebAudio or not.