Improve stall horn sound

[gilbertohasnofb]

improve stall horn sound, use this as reference: https://www.youtube.com/watch?v=4tLWIvnVnCQ#t=42

[gilbertohasnofb]

(as a reference, listen to this stall horn at 0:35 of this video. This is the sound I analysed for that synthesis: https://www.youtube.com/watch?v=zTq8Jvoks2U )

[onox]

@gilbertohasnofb Yours sound a little bit better, but both sound artificial. If we decide to use this, then we need to make sure we can accurately adjust the volume and pitch during a stall, so that it will sound like a screaming kitty :crying_cat_face: just like in those YouTube video's.

[onox]

Please create a branch bug-85 and a PR in that case.

[gilbertohasnofb]

both sound artificial

I agree completely, but that's the route that was decided. So now it's up to the magicians/programmers among us to implement this in a realistic way. I can of course cater to what is needed (e.g., I can create other variations of the stall sound, or a version that is cracking instead of constant, etc.).

Please create a branch bug-85 and a PR in that case.

Okay, I can do that, but from my part this will be only an upload of a wav file, nothing more. I really can't program in Nasal.

[onox]

Okay, I can do that, but from my part this will be only an upload of a wav file, nothing more. I really can't program in Nasal.

That's ok. As long as it's in a separate branch.

[gilbertohasnofb]

Okay, I will do that! Could you please tell me how to create a branch on my local repository?

[onox]

1. Check you're in master: git branch
2. git pull
3. git checkout -b bug-85
4. Add the .wav file

[gilbertohasnofb]

Okay, I did that all, then added, committed and pushed it, but I can't see the branch bug-85 on our repository using the GitHub webpage.

And by the way, the option -b on git checkout means what? Does it create a new branch in case it doesn't exist?

[onox]

-b means you create the branch and then do a checkout. git pull gives me nothing. Can you do git branch and tell me what you see?

[gilbertohasnofb]

I see:

andgi-add-water-sounds
    bug-238
    bug-80
* bug-85
    master

I was reading that I may need to push using git push -A -u instead of git push -A. Does that make sense to you?

[onox]

You need to do: git push -u origin bug-85

[gilbertohasnofb]

Done. Thanks a lot for the help, onox!

[onox]

Ok, now I get it:

remote: Counting objects: 4, done.
remote: Compressing objects: 100% (4/4), done.
remote: Total 4 (delta 0), reused 0 (delta 0), pack-reused 0
Unpacking objects: 100% (4/4), done.
From https://github.com/Juanvvc/c172p-detailed
 * [new branch]      bug-85     -> origin/bug-85
Already up-to-date.

Thanks! :+1:

[dany93]

@wkitty42 has written https://github.com/Juanvvc/c172p-detailed/issues/117#issuecomment-112890435

the idea is to find the pressures at the two points on the wing related to the stall horn... point A at the front (where you suck on the stall horn to test it) where the air exits the stall horn and point B where the high pressure air enters the stall horn's air scoop to be directed through the reed or valve assembly...

the idea is if we know the pressures at these two points, we can then calculate (or fake) the speed of the air through the reeds or valves... that speed will determine the pitch of the note..

if we can get qbar(??) from the front edge of the wing and from the bottom of the wing where the pressure is higher when in a stall, this might work out...

In JSBSim, qbar-psf is the dynamic pressure 1/2*ro*V^2. General value, not the detailed values at different locations. I think that these detailed value are not available (I do not know everything from JSBSim, if you find them, please tell me).

Even assuming that you have these differences,

we can then calculate (or fake) the speed of the air through the reeds or valves... that speed will determine the pitch of the note..

How?

As I've written here https://github.com/Juanvvc/c172p-detailed/issues/117#issuecomment-112883955 @dany93 wrote

I rather think you will have to use the AoA. It is the best indicator, admitted by everyone. By a non-linear relation if you prefer, you do all that you want. Anyway, at best imagine you have the pressure (difference), you have to apply a law to transform it into a note. And you ignore the physical law. Non linear for sure. The pitch and volume of the note are connected to the pressure by a law that you do not know, and the pressure is connected to the AoA, maybe to the velocity too. As the AoA is directly available, better to use the AoA. In any case you have to guess a law. You can add a velocity contribution if you wish, but with no better knowledge of the physical law. All of that is guess, or preferably comparison with IRL if someone can tell.

[onox]

@dany93 Let us say we have a function f(AoA) = (volume, pitch). We'll define how this function works precisely later. First question is: do we need to add more input variables to this function f?

[dany93]

Not easy to answer because (as I wrote above) we do not know which parameters have an effect on volume and pitch. The velocity may have an effect on the pressure difference. A stall can happen in straight and levelled flight conditions (that is around 45 KIAS) or dynamically, during a steep turn with high roll angle (higher velocity). Also at high g due to a strong pitch up. I completely ignore the possible effect on the volume and pitch. I think you will have to guess something based on the observations and subjective memory of c172 pilots. The first I would do is increasing volume (and pitch?) with the AoA. Then (maybe) volume with the velocity. Purely guess. Even for the pitch, I do not know why it changes. In a reed instrument, the pitch modulation is much more complex than only an air-pressure difference or velocity effect. In an accordion, the volume of the note depends on the pressure difference, but not the pitch (the note). In a oboe, the reed is in pursed (held) in the lips of the instrumentalist. Anyway, it can suffer some approximation with no important loss in realism....

[onox]

Currently the volume is controlled by /sim/alarms/stall-warning which is controlled in src/FDM/JSBSim/models/FGAerodynamics.cpp as follows:

161   if (alphaclmax != 0) {
162     if (in.Alpha > 0.85*alphaclmax) {
163       impending_stall = 10*(in.Alpha/alphaclmax - 0.85);
164     } else {
165       impending_stall = 0;
166     }
167   }

alphaclmax should be <aerodynamics><alphalimits><max> in c172p.xml, which is 0.28 rad.

[dany93]

That is: volume varies linearly with AoA only, with a threshold at Alpha/alphaclmax = 0.85. (true ?)

[onox]

So I think we could definitely write our own function that increases pitch when the AoA is in certain regions.

[onox]

What do you mean with volume?

[dany93]

If you want to include pitch, yes. "How to" remains to be seen.

The volume is the intensity of the sound, the power. Like the volume knob on an amplifier. The pitch is the frequency, the note itself (A, B, ... G). The volume can be increased by a continuous manner, but I do not call it pitch.

[onox]

I know.

[onox]

I think @wkitty42 posted some video somewhere that showed the different musical notes of the stall horn?

[onox]

@gilbertohasnofb Shall I update bug-85 to use a property which controls the pitch? You can then set this property to certain number (0.9, 1.0, 1.1, etc.) and give us a list of reasonable number.

[dany93]

At least gsagostinho on the forum http://forum.flightgear.org/viewtopic.php?f=4&t=25157&start=1995#p247508

[onox]

That's not the video I mean. It was a youtube link posted by @wkitty42

[onox]

This is the post I'm talking about: https://github.com/Juanvvc/c172p-detailed/issues/117#issuecomment-112844657 :smile:

[onox]

Can someone compute the angles where the pitch changes from that video?

[gilbertohasnofb]

Hi guys, Erik uploaded another version of his stall sound on this address http://www.adalin.com/ehtw.info/html/images/stall.wav . I personally still feel my version is a bit better, it sounds a bit rougher and gritier and more annoying (as it's supposed to be IMO). I also think that spectral analysis and resynthesis is the way to go instead of trying to approximate the sound via saw tooth generators and adding effects (it gets much closer to the sound we want to imitate). Anyway, I leave it up to you to see which sound you prefer. If you like his sound better, then I can exchange it myself in the repository, okay? Just please let me know.

[onox]

@gilbertohasnofb I think yours sounds better.

[onox]

Can I chop off that commit from @wlbragg in master? It gives the regular pontoons front gears :confused:

[gilbertohasnofb]

@onox @dany93 Erik posted a code snippet of how this could be implemented in our forum thread. Could you please have a look on it?

[onox]

I have force pushed master and bug-85.

[onox]

@gilbertohasnofb I'll take a look at the code snippet.

[gilbertohasnofb]

@onox thanks!

[onox]

Can you still take a look at the video in https://github.com/Juanvvc/c172p-detailed/issues/117#issuecomment-112844657 and try to estimate the difference in angles?

[gilbertohasnofb]

Yes I can do that but only a bit later, right now I have to leave. I will try to do it by tonight, okay?

[onox]

Ok.

[dany93]

I'd say an angle difference of about 8 to 10 deg between the beginning and the end.

[onox]

But currently the horn starts at 13.63 and ends at 16.04.

[dany93]

I find that it does not end. I had it up to high values. More than 20 - 22 deg at idle, 30 - 40 deg and more by climbing full throttle and cutting the throttle. It has no reason to end.

[onox]

Yeah, sorry, you're right about that.

[onox]

/sim/alarms/stall-warning is 0.0 at 85 % of the alphalimit and 1.5 at 100 %.

[onox]

I'm writing two <fcs_function>'s to control volume and pitch because I'm not too happy with the snippet posted by Erik.

[dany93]

For the code snippet, http://forum.flightgear.org/viewtopic.php?f=4&t=25157&start=2010#p247690 how do you understand it? I read 0.745 * log(alpha) Which would be a nonsense (alpha can take the value 0) And where is the threshold?

[onox]

This is what I have currently:

 277    <volume>
 278     <property>/fdm/jsbsim/sounds/stall-horn-volume</property>
 279    </volume>
 280    <pitch>
 281     <property>/fdm/jsbsim/sounds/stall-horn-pitch</property>
 282     <offset>0.0</offset>
 283    </pitch>

[onox]

Using a pitch between 0.9 and 1.1.

[dany93]

What is wrong with the original code https://github.com/Juanvvc/c172p-detailed/issues/85#issuecomment-113135864?

[onox]

The volume is linear.

[dany93]

And something like if alpha > alpha-threshold log(k * alpha/alpha-threshold) ? Linear at alphas close to alpha-threshold, then it tends to saturate with the log. The k factor enables setting the gain with alpha.