One-Synth-Challenge: General discussion thread

VST, AU, AAX, CLAP, etc. Plugin Virtual Instruments Discussion



Eauson wrote: Mon Apr 22, 2024 12:08 pm
Yadrichik_Chaya wrote: Mon Apr 22, 2024 8:22 am That jazzy house broken beat track that you made and then deleted was amazing, hope you reupload - had a blast jamming to it with the infamous Roland Jazz Scat preset :lol:
I'll have to look into the vaults ;)
Which synth, do you remember !?
Good question, I'm pretty sure it was one of the very last ones before you deleted stuff, just before, during or after Synister?
Or was it FB-7999? Around that time.

BTW I hope you've found good herbs/methods (like acupuncture) for handling migraines?

I used to have that shit very frequently when I was young from age 6-12 so I know very well how awful it is: sensitive to light, vomitting etc.

But in my case having had migraines turned out in my favour… :wink:

When I was forced to go and do tests for military service, me being a pacifist who has never been in a fight (and never will be) or would even harm an insect, when seeing the doctor there I had some tricks up my sleeve. :clown:

First, since I'm very agile and can ”bend” back my knees (looks a bit freaky) the doctor really thought there was something wrong with my legs… :lol:

He then said that I could still, despite my legs, work in the kitchen but I replied that unfortunately I have frequent migraines and mentioned the name of some migraine medication! :phew:

Problem solved, no military service for me in any shape or form. :tu:

Sure I might have ”stretched” the truth there a little (pun intended), but no f**king way that I would want anything to do with violence/war/weapons - but despite this I have actually trained martial arts, but never for actual fighting - that is why I suggest you look up stuff like Tai Chi & Qi Gong exercises, these exercises will get the ”Qi flowing” and prevent migraines. Just google it, you'll most likely find a lot of stuff that actually works.

Or you could get some CBG buds to smoke/vape or CBG oil/isolate.
CBG is amazing for all types of things, love it! :love:
Migraines: CBG appears to act as a vasodilator with strong neuroprotective characteristics which might make it an interesting choice for migraine prevention or treatment because the severe and often debilitating headaches occur because of vascular dilation. 


Yadrichik_Chaya wrote: Wed Apr 24, 2024 3:56 am
I take a look date wise & see what's what ::)
Yes, CBD resine is helping with additions, has done since my youth to help with a myriad of problems ;)
Man is least himself when he talks in the first person. Give him a mask, and he'll show you his true face


Eauson wrote: Wed Apr 24, 2024 5:09 am
Yadrichik_Chaya wrote: Wed Apr 24, 2024 3:56 am
I take a look date wise & see what's what ::)
Yes, CBD resine is helping with additions, has done since my youth to help with a myriad of problems ;)
Hope to hear the track again! :tu:
Yes CBD, CBG, CBN etc. all of them really amazing. :party:
(I haven't tried THCV, but plan to… :wink: )


Sorry for the noob question - but what exactly is granular delay/reverb and why is it not permitted as part of the rules?


abot81 wrote: Thu Apr 25, 2024 3:01 am Sorry for the noob question - but what exactly is granular delay/reverb and why is it not permitted as part of the rules?
I assume that granular reverbs and delays chop up the incoming signal into grains, then processes and spreads them out in unique sounding, evolving ways that conventional reverbs cannot do. It's probably just too much of a potential diversion from the original source sound. I believe that some granular processors tend to add some pitch shifting and noticable modulation effects too.


Yadrichik_Chaya wrote: Wed Apr 24, 2024 5:24 am
dL enabled
Man is least himself when he talks in the first person. Give him a mask, and he'll show you his true face


Shameless self promo ... Silverpants has asked a few weeks back why I submit my tracks so early at the beginning of the month. I replied that I'm working on several tracks parallel to OSC and need to stop starting but start finishing ... this is actually one of the parallel tracks which I now call finished ;-) ... echno-club


From the Charlatan 3 forum:
Impedance Music wrote: Thu May 16, 2024 9:06 am
doctorbob wrote: Wed May 15, 2024 12:33 pm Which software did you use to convert the audio to MIDI? Various software comes to mind, and even simple ReaTune in Reaper can do it ...

Just wondered which way you went with this ...

I've used this free Online Converter to create the midi:

FrogsInPants wrote: Thu May 16, 2024 7:22 am
silverpants wrote: Thu May 16, 2024 3:22 am I think what Impedance has done is really fun and inventive and I have no problem with it.

However - it does make me think about where this method could end - and that could be outside the spirit of OSC. [...] It wouldn't break the rules of OSC, but I think it would break the spirit.
I appreciate this as an experiment, but I think it's already pushing the boundaries of the rules. [...] No use of samples is allowed. Arguably, that applies not only to use directly as a sound source, but also to side chaining and computing parameter inputs. If not, perhaps it should, so we have some kind of clear line on the matter. :shrug:
Having read your comments: Yes, i agree - this is pretty much on the edge of what i would consider fine with the rules. An additional rule clarifying that computed inputs of actual audio samples are not permitted sounds good to me. Especially because it didn't occur to me as using a sample when creating the track.
Might be a good idea to discuss this, and perhaps the possible impacts of AI for the challenge here, with a view to possible modifications of the rules.


Oops wrong tab :lol:
"We should not kill the earth, we wouldn´t know where to bury it!" - Michel Granger


For what it's worth, I agree with what was said that generally speaking, using waveform audio as any type of input to the synth should not be allowed. In this case, it's MIDI input that was generated from waveform audio used as note input. But of course, that would also apply to using waveforms to modulate parameters and also using waveforms to generate patches. This should be true for all waveform data: existing or generated, in my opinion, to be truest to what I think the spirit of the competition is.

There's another thing to consider: wavetable synths, and especially: Vital text-to-wavetable (TTWT). This, specifically, has been allowed in the past and maybe still should be because it's a function of the synth rather than waveform input to it, but that could preclude using patches created with TTWT. And similarly, if you create wavetables (or even use them!) from audio, this could be disallowed but it would eliminate a set of things that I think fall squarely into the spirit of the competition like making true wavetables for a piano sound, *ahem*. For me, it's hard to qualify exactly what the "spirit" is in some of these cases.


At the moment we are allowed to generate MIDI data with external programs, import that to a track, and then render it with the synth.

I would think that generating MIDI from speech or singing (ie audio), is, I suggest, perhaps not too far from the above. Also, I have a hardware audio->midi box which has voice or guitar input, and outputs MIDI in real time. Would the MIDI from that be allowed?

The discussion about speech to MIDI is an interesting intellectual argument in that as I said, your would have to track the formants of the utterance (think why we have such things as formant filters etc!). You would also have to emulate fricatives and plosives, which might be OK to do with most synths. The underlying thing is the basic waveform generator required - we would need an oscillator to generate the right form of glottal pulses - a wavetable synth could do that.

Programming the F0 (equates to fundamental) would then allow singing.

Additional formants F4,F5 would be needed for higher voices and female voices. Also, women, in general have shorter vocal tracts, so again this would need to be allowed for.

Finally, some form of emulation of the nasal resonance is needed (not to sound nasal!) in order to get more towards the human speech.

The model of how a midi stack could look for e.g. a syllable, would be:

F0,F1,F3 for the full duration of the vowel sound in the syllable, then the other elements are layered on top - eg CAT would have the full sound of "A", and the "C" and "T" sounds layered over that. So COT, would be the same except the underlying vowel would the "O".

In fact, the whole vocal tract configuration is set for the vowel before we utter the word. Can't say "beet" whilst configuring for "boot"!!! Everything is in the wrong place!!!

But, back to the discussion. If we want to craft MIDI to sound like speech, why not? To get something sounding OK in the mix, that's relatively easy, but to get say some acapella, a different story altogether.



The problem to me is that at least in theory, it's a lot easier than you've described. Essentially all you need is a synth that plays sinewaves. You run the audio through a windowed FFT, and get a bunch of MIDI notes corresponding to the exact times certain frequencies are present in the audio. With enough resolution, when this is played back you could theoretically exactly replicate that waveform. So even though you're not directly playing a sample, this is effectively getting sample playback.

I'm pretty sure at some point also there either was a mention in the rules or a post on KVR here regarding wavetable synths: that it's not allowed to use the wavetable oscillators to emulate sample playback, but I don't see this now. But the rules do clearly state that samples are not allowed. So where does that line live? To me, if you get something sounding like the waveform by using the waveform, not by tweaking envelopes and settings manually, this straddles the line a bit too far.

That's not to say don't try to get speech, just don't do something that is an attempt to get waveform playback. To me, if this is allowed, we should also be able to modulate synth parameters using waveform data. I'm not opposed to changing the spirit of the OSC, but I just don't see "working around sample playback rules by way of MIDI notes" as something that's in the spirit of the current set of rules.


Interesting to note that there is a US patent US5915237A for inventing speech to a MIDI representation ...

You are right about basically extracting the formant frequencies over time from the fft's and then getting midi notes (and maybe pitch bend params to fit small changes) and using eg sine aves to replay those notes, essentially reproducing the original audio signal - I see this does stretch the rules, and could become "sample playback", just not as we generally know it!

The methodology I was describing was how to create the human-like speech with a speech synthesiser (been there, done that!). Getting it to sound natural in the text->speech domain is still difficult (e.g. Stephen Hawking box, which has KLATT at the bottom end, just driven with clever stuff to get phrasing and intonation into it). The original code has LOADS of little "fixes" in it to allow for strange pronunciation conditions and also some words in combination need to be pronounced differently. Basically, there wasn't a complete model of human speech, so could not be a generic synth.

We vibrate our vocal folds, and that makes a glottal pulse, which pushes the air up to the various head cavities, which we then control to make speech sounds. Sine waves are OK for reproduction, but won't work for creating natural sounding speech, you need the pulse clearly seen in the spectrograms of actual speech. Also, even if a synth makes natural sounding speech, it can be exposed as fake by analysing the glottal pulses.

I guess we can still try to create speech synthesis via MIDI, without pushing it towards sample playback. At the moment I would think the jury is out on whether it pushes the boundaries too far.



If DoctorBob talks about using formant filters built into the synthesizer with modulated (inside the synthesizer) noise and tonal sounds, this should be fine. In any case, this is a very time-consuming method and if someone gets a speech in this way, it will be impressive!

As for the audio to midi, it uses a psychoacoustic effect, almost making us hear what is not there, but I agree with Z'Prime that in spirit it is a sample playback. The rule about modulation at the speed of sound works well, and this is one of the (very specific) violations of this rule. Porbably some people here can play more than 20 notes per second, but obviously not for the purpose of reproducing phantom sounds) No matter how much we joke about manually twisting the knobs at sonic speed, our natural physiological limitations are one of the most reliable rules!

Wavetables is one of the most gray zones fulled themes. These are tables that reproduce samples. These are creating your own tables using mathematical formulas. We need to come back to this topic every time a new WT synthesizer pushes the borders. Now I think that the fairest and in the spirit of OSC would be to allow the use of all synthesizer abilities and ban any external sources. I.e. if there are piano tables in Vaporizer, then this is allowed for Vaporizer in any way, but no external tables. If some synthesizer has a sampler with built-in samples, then their use is allowed, but no external samples. For external wavetables, this is either a complete ban or the creation of an unified list of publicly available tables of simple signals that we can upload to each OSC WT synthesizer. After all, it is sometimes interesting to work with broader possibilities than classical synthesis methods, the main thing for OSC is that everyone remains on an equal footing.

P.S. I do not consider Impedance Music to be a violator this month, on the contrary, I admire his achievement and am glad that he raised an important topic for OSC!

P.P.S. And I would like Richard Semper to explain in more detail the essence of his proposed discussion of AI in the context of OSC. Since, as I understand it, the audio to midi is made without AI, so it looks like this is a separate topic.

Return to “Instruments”