Hi there!
I am working on a metronome for Android. When I run the patch in PureData-editor all works fine, sound is even. But when I run the patch thru libpd on phone the sound is not even, the volume is some periods a bit weaker than other, let say 3-4 beats the sound is strong and even, then the sound gets weaker for 3-4 beats.....and then it all repeats.

Any thoughts that could help is welcome.

Same behavior with another unit....

However, I changed the wav files that was loaded in patch into mono instead of stereo and it seemed to make a change!

But, Im maybe getting crazy after listening on the sounds, I imagine I hear a straight attack on the sounds some time and another time there is a "longer" attack.

/Thomas

@TompaD I understand "getting crazy listening"........... Our brains always do "weird stuff" with what goes into our ears.........
Is your Android app something that you could post for testing by the community?....... or will it be a paid app?....... (and so you don't want to post it here?).......
If you do want to post it you might get more feedback............
David.

Yes, maybe I will make it a paid app, but I realy need to get the sound
good (and a lot of other things). Right now I building it for my self, I'm
my own beta-tester

I've never worked with mobile devices, so I don't know if it works the same, but I don't see anything in your patch to turn DSP on. Could this be the problem?

If not, my advice is to start deleting (in a new file) bits of the patch one by one and testing it, until you find the minimum configuration that displays the behaviour you're describing. Is it the table playback that is failing? Can you get a simple [osc~]--[dac~] configuration to generate sound? Once you've narrowed it down, maybe you can share your patch and then we'll be able to help.

Hi LiamG!

Thanks for your response. The patch IS working in PureData-editor
and on my mobile thru libpd, However on my mobile I get some
unwanted behavior of the sound (see earlier posts). Correct me if
I am wrong, but if I get sound from patch/libpd the dsp must be
turned on? Or?

/Thomas

You can init the dsp from your third party code, or in your pd patch. It doesn't really matter.

Do you use mobmuplat, pd-party, scene-player or your own code to load and init the patch ? I think some of those init the dsp and some don't.

Anyhow you can always use in your patch to activate dsp.
[loadbang]
|
[; pd dsp 1; (

Did you try normalizing your samples on import ? read -resize -normalize.

When on mobile I often use a soft clipping abstraction before the output, somehting like [tanh~] that you can probably find easily on this forum.

There is also an abstraction called [e_agc] in the rjlib : https://github.com/rjdj/rjlib/tree/master/rj
it's an automatic gain control, it could help.

Hi Berenger!

I am using my "own" code...from Peter Brinkmanns book "Making Musical Apps".

Hmmm intresting! -normalize have I not used yet. Will check it out.

Soft clipping abstraction - I am aware of soft clipping, but what is soft clipping abstraction??

tanh~ - is this part of "standard" puredata (not extended)?
My editor couldnt find it.....

Very interesting input! Thank you Berenger!

@TompaD Bored and googling....... I had thought "noise suppression"....... unlikely, and "echo cancellation"....... very likely, but only available for audio acquisition on android.
So the only thing left (I think) is "duck"...... but another app would have to be repeatedly demanding and relinquishing focus. That is possible? I suppose so............
Do you have AUDIOFOCUS_LOSS_TRANSIENT_CAN_DUCK in your code?

Also, the samples do look quite low level, so as @Berenger said...using the -normalize switch with each "read" will help the volume by making the sample as loud as possible without distortion....
ALSO...... maybe you should post your original question here........ http://stackoverflow.com/questions/tagged/libpd?sort=newest&pagesize=30
David.

for tanh~ it's not really extended but it's not an object of vanilla either

you can do one with expr~ : [expr~ tanh($v1)]

or you can use table lookup : https://github.com/chr15m/blockhead/blob/master/e_tanh.pd

@whale-av Thanks for the tip's!

I am not using AUDIOFOCUS_LOSS_TRANSIENT_CAN_DUCK, not what I know.
Should I? ....woav, found a lot info about it on Google, need to read, will be back.

The -normalize switch is set, but I dont know if it did any change, hard to tell.

@Berenger ?? Where can I find newbie information about tanh~??
What does it do?
I visited github link, but I am to green to understand that information.

If it's any help, [overdrive~] in the Cyclone library does (I think) the same thing as tanh~.

@TompaD, you should go there instead : https://github.com/chr15m/blockhead

click download as zip and checkout the help patch, it shows the effect of tanh~ at high volumes by drawing the signal in a table.

a soft clipper is like clip~ but with rounded 'edges' : with clip you cut clean everything that over 1 and below -1 (for instance) with a soft clipper when you reach those maximums softly with a little interpolation. It's like a limiter with a knee.

BTW did you try with headphones on and do you have the same issue with headphones ?

I aware that you don't do algorave music but this could be of interest to you too :
https://mccormick.cx/news/entries/heuristic-for-algorave-mastering

@Berenger said:

a soft clipper is like clip~ but with rounded 'edges' : with clip you cut clean everything that over 1 and below -1 (for instance) with a soft clipper when you reach those maximums softly with a little interpolation. It's like a limiter with a knee.

Aah now we are talking! Limiter I know!

BTW did you try with headphones on and do you have the same issue with headphones ?

I tried with headphones and I think it was not the same, so it is something with the output to the speaker....

I also made normalization on the wavs in my soundeditor and removed the -normalize in pd patch. Now the sound is a bit higher in volume, I dont need to put the output on phone more than ca 50% to hear it clear.....we are making progress!