@Nicolas-Danet said:

"The idea behind Spaghettis" is to clean the code of Pure Data in order to improve it.

This is a noble task . Have you considered sharing those improvments to Pd, Purr Data or LibPd with merge requests? Why not contributing to those projects instead? Not criticising, just asking

Thanks a lot!

While waiting for the Pd 0.49 stable release and Purr data 2.6, I found pd-faust. How does faustgen~ differ from this external?
On another topic, Purr data is now supporting 64bits floats. As Faust can already do this, is there any particular thing to set in faustgen~ so it is aware of this new feature?

This is great! Thanks a lot!
I havn't tried it yet and won't be able to do so in the next few weeks but as far as I know, Faust can't work with multiple sample rate. Is it possible to circumvent this by putting a faustgen~ into a subpatch with a different sample rate?
Like doing 8x antialiasing on a faust filter by upsampling/filtering/downsampling?

Great, thanks a lot!

Sorry to bother you again, @nau, but, if you still have it could you please post your patch here, the sendspace link is dead

@whale-av
Well, I didn't think of using [env~] this way but theoretically this could work (and be computationally expensive) if:

• I set [block~] in a way that it produce 64 blocks sent per block received, with each block sent delayed by one sample. Would that be [block~ 64 64 1], [block~ 64 64 64] or [block 4096 64 64]?
• [env~ 64] could then process the average of those 1 block/64 samples received and at the same time keep up at sample rate as it is under heavy fire then I could use [snapshot~] to get the polarity right, as [env~] only sends [0-1]. Not sure if this would be nice to hear though.

Anyway, I think it's time for me to let go the rule I set for myself about vanilla only/no externals as there is exactly what I need in cyclone with [buffir~]

Thanks for your insight!

Hi!
I would like to create an averaging filter that takes the last x samples and output their mean/average value.
Can this be done in vanilla?
Would the overlap factor of block~/switch~ be usefull? (What is its use generally speaking?)
Is there a way to run this filter with the use of a gaussian or another arbitrary curve to calculate the average output?
Any help welcomed!

@whale-av , I knew about Camomille but thought of something different from the words used ^^'

@EEight said:

[...], so that it is possible to make plug-ins of various sorts out of Pd and they can coexist with each other. This will allow people to write much better VST plug-ins in Pd.

Maybe I've missed something but how can you "write VST plugins in PD"? Is there a tool to convert PD patches into native VST plugins?

[mtof] and [mtof~] convert MIDI pitch to frequency so it's easier to deal with transposition. That way you process the pitch using integers (add 1 and you go up a semitone, subtract 12 and you are an octave lower), send the result to an [mtof] and let it generate the frequency for oscillator.

Sorry for the double post, I realized that this topic should have been posted in the patch section. Can a mod move it there please?

Hi!

This is not a perfect emulation as the TB-303 has unique oscillators and filter designs. I tried my best with simple bandlimited oscillators and two cascaded bob~ and focused on modulation behavior instead. As I don't have a 303 at home, I used Propellerheads Rebirth RB-338 to match (or rather guesstimate) settings by ear.
Design based on an analysis of the TB-303 by Robin Whittle, Oscillators by @Maelstorm

I can sounds somehow convincing if:

• you drive it with a sequencer using 1/32 note length with 1/32t silence gap between them. Two notes superposed will trigger the legato. Use velocity >= 100 for accented notes.
• you send it to a distortion unit (Guitarix is perfect for that).

Have fun, any feedback is welcomed!

YLB-303.zip

Yes! Thanks a lot!

I mean I can already plot the curve, I was just asking if there was a way to apply the frequency response described by that curve to the signal.

Thanks for the help! I'll try that way, it's seems quite easy.
One last question, let's say that the signal goes trough multiple filters (one low pass and then 2 high pass with different settings). Is it possible to generate a curve by sweeping a sine from 20Hz to 20kHz and apply that curve as a filter in the same way you can use impulse response for reverbs?

Indeed it looks like one, the thing I'm missing is the method to get those parameters
How can I translate the peak frequency, gain @ peak frequency and slope into those parameters?

Hi! I'm playing with electronic circuits simulation (with LTSpice IV) and can get interesting curves like this one.

I of course also get a csv file containing the numbers that are used to display this curve.
Is there a simple way to deduce biquad filter parameters from that?

Hi!

I found this really interesting thesis on the Roland Supersaw emulation and tried to get it done in Pd.
Unfortunately, there is a part where it is needed to manipulate large numbers like 17019.9518580080 (see page 12) and these can't be handled by Pd vanilla. I tried to round them but the result isn't what is expected. I also tried [expr] and this gave worst results.
So apparently, I have to rely on externals to get this done. Coding one in C is out of reach for me for the now. Would PDlua be able to handle those numbers?

I use checkinstall whenever I can, I find this easier for keeping a clean system