New Core Technology and Additive Synthesis

I was wondering, (please forgive my lack of knowlege with this), as to whether the new core technology will expand Reaktor's potential in additive synthesis. It seems like the module and macro department is somewhat lacking in this area, and the impression that I get from the users library is that additive synthesis in emsembles is a rather new phenomenon.

I've been working my way through the entire users manual, and have come across one oscillator module for additive synthesis. I've yet to reach the core manual, when I do, I intend to go through the whole thing.

Might it be possible, using core technology, to create new core cells specifically for use in additive synthesis? I'll probably not have much to say in this conversation, as I'm fairly new at this.... I've sampled some of the additive offerings in the user library, (Fourious definitely stands out, but it can be problematic at times, with snapshots sometimes not firing up). The reason I'm asking is that I would like to use additive synthesis more in my work, and I'm curious as to whether core can expand this area.

 

yes it is possible and easier than in primary level.
F.i. i have created a trautonium osc (see hardware here: trautonium osc) in core. An older version in primary level you can download from the user library mixtur trautonium.
A good starting point is the core osc "SUB OSC 4" which is made very tricky.
And very impressive is the WELLENGEBER by Gabriel Mulzer.

ciao herw

 

Fourious is well made but the best are the snaps. They are very well programmed.

ciao herw

 

Mixtur Trautonium is excellent! I was checking out it's structure, it definitely produces some ideas. It reminds me a lot of my beloved arp odyssey in terms of sound, as a matter of fact it gets closer to that than any softsynth I've heard so far. I do hope to be able to post an additive synth, but I've got a lot of work to do. You've definitely got me thinking.

It would seem to me that core technology might enable one to push the envelope in terms of additive synthesis. I'm hoping that it will allow for more flexibility in the generator than primary.

 

I love Fourious, but I've had some trouble getting the snaps that I've programmed to start up. With one, I had to hit a note on the origional snap, and switch over to the new patch while the previous note is still ringing. Then it seems to work fine.

 

theres a sinebank module, in the buildingblocks macros, made with core, but its not particularly userfriendly, and is set in stone @ 16 sinewaves unless you add more. its still better than the 4-osc multisine module. i personally love the trautonium osc myself (from corecell library and subharmonic), and am looking forward to herw's new ens.

 

)

ciao herw

(just playing and patching with mm2)

 

Please excuse my confusion, but I was unable to find the trautonium oscillator in the core cell library, are you speaking of the corecell sub osc, as used in Subharmonic?

 

Ooopps. I forgot about the jpg of the oscillator that herw posted. While my german is nonexistent, it seems to utilize 4 sub osc corecells? (I'm going from memory).

 

you can not find the trautonium osc in the NI-library because it is my own creation and still not published. It will be published in next weeks as soon as i upload the MODULAR TRAUTONIUM.
You are right its main structure are the four Mst oscillators. That's nothing new. The new idea is to get their pitch from one input by transforming to frequency and calculating the increment of every sub oscillator. I have splitted the Mst osc in two modules (sub inc and 4*subosc.). The main difference to adding and mixing normal oscillators is, that the four oscillators are in phase. That's typically for the trautonium oscillator. But that's not the only difference. The original trautonium has formant filters which means that they do not depend from the pitch and have fixed cutoff frequencies. Therefore the trautonium sounds sometimes like humans voice vocals aaah, oooh etc.
I have added some mp3 (sorry for bad quality but there is the 100kB limit).
To start your own additive ensemble it is a nice idea to use the formant oscillators from the NI library. I will use them in my MODULAR EINS ensemble (uploading september 2006 i think).
NI-library shows another way with the sub osc 4 module There are four sinus oscillators in phase.
It is interesting to understand the idea of building oscillators with core cells.
ciao herw

 

I very much look forward to that upload! My own ensemble is still in the research phase, but I can see the usefulness of formant oscillation. It's no accident that it was used in Subharmonic. It's going to be a while as I've decided to read through every module and corecell in Reaktor before even starting. The biggest problem is going to be keeping the audio pathways to a minimum required to do the job, or it will be an unacceptable cpu hog. Even a cursory look at additive synthesis and it's requirements shows that cpu usage will be an issue.

 

IMO, additive synthesis is really useful when there is an analysis stage attached to it, like in chameleon 2000. this is a good way to balance the power of additive synthesis with the mind-boggling amount of choices and parameters. and this is the tricky part in Reaktor, as stuff like this is probably easier to do in something taylor made for it. not that it cant be done in REaktor, people like Gabriel Mulzer already showed us a few times how FFT analysis/resynthesis can be done in R4 and R5, but getting it past the proof-of-concept phase is not a trivial thing. but if you are up to doing anything in this direction, you definately have my support.

 

the primary multi sine osc, as far as I recall, doesnt allow for arbitrary frequencies, only integer multiples of the base fq, other fq's are truncated at that rate, this and some other things make it pretty unusable

right now I am working on an FFT offline analysis tool, that extracts major partials, exact frequencies and a spectral envelopes, it can also resynthesize a matching waveset for waveset synthesis , maybe it will also create a formant estimation, I'll post some pictures
This ens improves on similar stuff I've posted previously ( like the now- delete Anna-Lisa ensembles ) but still needs some more work it uses a 4096 size FFT which seems to be the best match for this actually I didnt think of posting it, not sure if this is going to be useful for anybody, but if there is interest I'll post it - provided I find the time to clean it up etc, there are lot's of similar things on my HD that could be useful but I dont find the time to put them in a postable condition and to add essential description and information...

FFT synthesis - I am still encountering problems with that I am not sure if it's a logical problem ( adding wrong sections at a the wrong time ), or a problem with the frequncy tracking, or already working as expected unfortunately, all of the above produce the same artefacts so I cant use my ears to find the problem, and I right now am absorbed by something completely different, and just needed the FFT analysis for something that.. so I still have no idea how feasible FFT for synthesis would be and wether its possible to have several voices or not and how much CPU is left for any smart controls and processing

Resynthesis with sines or sines + noise :
this depends very much on the source material - in some cases down to 16-20 partials and less seem to give quite good results, and the method has the advantage of zero latency and less temporal blurring on the resynthesis - but usually, much more sines are needed

Synthesis via wavesets : has the disadvantage that it sounds rather static and artificial due to the

restriction on integer harmonics, but still can be useful since it needs the least amount of CPU to play back

if it makes sense to use the core depends from case to case and on the task CPU will be a problem either way

 

It will be a LONG haul. You're right, getting from concept to actuality is truly a battle. I intend to develop it over time. I'm a musician foremost, so I intend to go slow. And I'm sure I'll need that support!