Latest firmware : 2.10 (2019 june 13th)

[Xavier]

Thanks for the suggestions.

Algorithm and oscillator shape controled by matrix would lead to audio glitch (in a bad way such as digital "click") , so i don't think it's a good idea to have it as part of the "out of the box" engine.
Triggering the enveloppe with other LFO value is not easy.

And i admit having all free envelope parameters would be cool 

[Xavier]

A little new release 2.10 is available :
https://github.com/Ixox/preenfm2/releases

[audiowanderer]

Just checked the latest firmware and goes pretty sweet.
I had some problems saving and accessing via usb cable to the usb stick with the 2.08b and now those problems are gone.
Great job.

[kyoto]

Xavier, thnx very much for your work and great fm maschine

is it possible to build in the firmware new options please:
1) master tune parameter with range for example (428 - 450 Hz)?
2) different types of LP filter (12db/18db/24db)
3) more user waves for operators (now the default is only 6)
4) Do you plan to add to the preenfm2 editor work with banks of sounds and patches?

[Xavier]

Hi thanks for the nice words and thanks for the suggestions.

I'm afraid the replies to your questions are :

1. Yes but you can already do that with the scala file (not ideal i agree).
http://www.huygens-fokker.org/scala/scl_format.html
I think someone already made a custom firmware with this. I would need to merge it.

2. I don't think i'll ever do that. But who knows. FM synthesis should not rely on filters even it's it's a good to have.

3. No it's limited by the memory of the preenfm2 which is already full.

4. No, that would be too much work.

[kyoto]

1) a little awkward 
i saw on other topic custom extended firmware from lokki, his firmware has master tune parameter, but only with range 435 to 444.
http://ixox.fr/forum/index.php?topic=69326.msg72771#msg72771
I need a lower mastertune frequency, since I work with live musicians who tune their instruments for eample in 428hz and 432hz

2) the LP filter that is now in the firmware 2.10 sounds very dirty

it's easy to hear:
reduce the resonance parameter to zero
and gradually reduce the filter parameter
I hear that resonance still affects the sound

I understand that the filter is not the main thing that should decorate this synthesizer 
but sometimes when creating patches of high frequencies there is a lot of and I would like to soften the sound a bit
I don't want to use external filters or vst plug-ins for this, I would like to have a good LP filter in the synthesizer itself 

3) so we store user waves on the external storage, and there is enough space for storage

4) I hope in some of the releases you will do work with banks / patches

thanks anyway for the work done!

[Xavier]

Thanks for your interest

2) I'm sure the LP filter is not very good as it's a simple formula (CPU light) i found on the web. A better one would be heavier and could unlikely be handled by the preenfm2 CPU without sacrifying some voices.
Writting good filter algorithms requires time and skills i don't have.
But if you find some opensource interesting ones, i can take a look.

3) All waveforms must loaded in RAM at runtime, so that the CPU can access them efficiently. RAM is much faster than Flash storage.
The RAM (192kb) is quite full. UserWaveform takes 24kb of them.



Xavier

[Toltekradiation]

hi!
I would also love to have some more filter flavour on the pfm2 , so here is what i found :

a biquad implementation, in C, with LPF,  HPF,  BPF, NOTCH, PEQ, LSH,  HSH
https://www.musicdsp.org/en/latest/Filters/64-biquad-c-code.html

a 3 band equalizer
https://www.musicdsp.org/en/latest/Filters/236-3-band-equaliser.html

a moog ladder lightweight approximation, with lp, bp, hp output :
this one may be really great, even in 1 or 2 pole mode : mixing the 3 output may be also interesting...
https://www.musicdsp.org/en/latest/Filters/145-stilson-s-moog-filter-code.html

finger crossed 

[Xavier]

Thanks a lot for these interesting links.

I doubt any of them would fit the required low CPU usage.
Once the preenfm plays 8 voices of 6 ops algos, the CPU left i very small.

It would also takes time (i don't currenly have) to implement them.

If anyone wants to give a try, i can guide you through the code files.

[Toltekradiation]

ok, thanks, i may try it myself ; i found the filter code part in the git repo. I'm not particularly fluent in C but it seem doable, so we'll see. I'll let you know if i come across something interesting.
cheers !

[kyoto]

to Toltekradiation - wow - great ideas - I look forward to and I hope you succeed
Hooray

[SMF]

Hi,

I'm not too sure of this one but could it be that I found a bug/glitch in the pitch-wheel-processing of 2.10? Pitch-bend-messages seem to be not followed in the full range.

What I expected:
Midi-Pitch-wheel-Messages are followed in the range of +/- 8191 values.

What I found:
Midi-Pitch-wheel-Messages are followed in  the range of (aprox.) +/- 1280 values.
There is some uncertainty in this value -- therefore the "approx." -- because I have tested this from within a DAW and "by ear" until there was no audible change any more.

How to reproduce:
Create patch mapping pitch-bend to all-osc-freq (or all-osc-freq-harm) and create a pitch-bend-transition with your favourite DAW in the range of +/-8191.

best,
Stefan