Tone2 Rayblaster: OUT NOW! (demo version available)

[Sendy]

Not to blow my own trumpet, but my analysis I posted a few pages back I think explains what's going on fairly well in terms of granular synthesis. The crux is that pitch is handled by retriggering at the desired frequency, and the spectrum of the waveform is shifted up and down be depitching the wave. You can then apply various types of mixing for PW effects and of course add an envelope to smooth out the transitions. There's more to it but that is the basis of how RB decouples timbre from pitch.

On the one hand, one waveform would be hard pushed to cover a whole filter's range from sub bass to ultra treble, plus you won't get the range of tones and nonlinearities a real filter would have. But on the other hand, it can be very effective - for example the MS-20 presets I thought were very evocative - so not only does it work but it's something DIFFERENT and that's one of the reasons I like it.

[Ingonator]

Basically you are sampling a saw waveform (for example) that has been filtered. You then cut out a single cycle of this sample. RB then magically manges to extract a filter from that?

BTW, what about pitch, how is RB handling this? what pitch are you sampling at?

Yes, that's basically it. You then then look that you got a zero-crossing at the beginning and the end and if possible that the part with the biggest amplitude is at the beginning which does not work in all cases. Anyway you could still adjust the waveform with e.g. different "osc windows", harmonic value and the start phase parameter later.
Sampling and using different basic waveshapes (e.g. Square, Triangle) does work but with the "PW sequences" Rayblaster could do some nice emulations of different waveshapes so a Sawtooth is usually enough.

About pitch:
It good to sample at a very low pitch for a good low end. Usually i use e.g. C1 or C0 for the samples/waveforms. You could also get results with higher notes but in the low frequencies it will sound a bit thin or "unnatural".

As already mentioned a basic value for the Cutoff in the original synth is around 25% to 40% (around 30% in average). The percentage is simply based on the knob/slider range and not on the real filter frequency.
As already mentioned a basic rule is to have a Cutoff at around 1000 Hz or lower (at a Signal Analyzer). This will result in the biggest range for the Formant knob where the usual range is between -200 (or a bit bigger) and 0 and then will slowly change into a Bandpass filter. You could see that at the PDF of the Saurus example which i posted a few post above.
For very high and very low frequencies it could be necessary to have additional waveforms for e.g. full Cutoff and 50% Cutoff (= middle position of the knob/slider). If you are not sure could do steps like e.g. 20%, 30%, 40%, 50% and 100% (= full Cutoff) and the check it in Rayblaster later.


Ingo

[hakey]

There's more to it but that is the basis of how RB decouples timbre from pitch.

Decoupling timbre and pitch, so that, eg, a vocal sample can be repitched without changing its duration, is what resynthesis is about.

But that has little to do with capturing a filter response.

Different concepts altogether.

[highkoo]

And the thread gets weirder!
Dunno Urs, you could try and rise above, and just be flattered, in this case.

With a synth like this, and mystery surrounding its workings, lots of 'testing' is going to be done to try and nail it down.
Im not looking for a discussion, but consider that the Diva test took place because Diva is at the top of the heap, and for no other reason?

Id expect many many tests to be done. Not necessarily as a competition, just as a way to figure out what RB does via comparisons.

Im interested in what hardware samples do in there....
I think I dumped that ol' SlimPhatty wav in there for a minute, as is my usual.
I dont feel that I have enough quality hardware ss wavs though...
Nor will I ever.

[hakey]

From the manual:

Rayblaster can take two cross blend-able samples, loops, waveforms or even an impulse response of any filter, to use as source for its IMS oscillators.

So it can use filter impulse responses.

[hakey]

...but further on it explains how to make your own filter impulse response:

1) Open any synth
2) Play a low pitch sawtooth (around 50 Hz) with any filter (cutoff around 1kHz, you
can add some resonance if you want)
3) Record and save this static sound to a wav.
4) Cut edit this wav into a single-cycle waveform.
5) Load it into one of Rayblaster's oscillator.

I guess some people have take this to mean that by following this process it's possible to capture the behaviour of the source synth's filter!

[Ingonator]

...but further on it explains how to make your own filter impulse response:

1) Open any synth
2) Play a low pitch sawtooth (around 50 Hz) with any filter (cutoff around 1kHz, you
can add some resonance if you want)
3) Record and save this static sound to a wav.
4) Cut edit this wav into a single-cycle waveform.
5) Load it into one of Rayblaster's oscillator.

I guess some people have take this to mean that by following this process it's possible to capture the behaviour of the source synth's filter!

But that's actually the result of that procedure. You get a waveform that captures the filter response of the filter and Rayblaster could extract that filter response from the waveform.


Ingo

[hakey]

You get a waveform that captures the filter response of the filter and Rayblaster could extract that filter response from the waveform.

But this process can only capture a one dimensional image of the filter, namely its response at whatever frequency/resonance/input-gain is chosen - it can't reproduce anything like its full range of behaviour.

Someone correct me if I'm wrong, but wouldn't a filter sweep yield a more accurate (though still limited) impulse response?

[pdxindy]

You get a waveform that captures the filter response of the filter and Rayblaster could extract that filter response from the waveform.

But this process can only capture a one dimensional image of the filter, namely its response at whatever frequency/resonance/input-gain is chosen - it can't reproduce anything like its full range of behaviour.

Someone correct me if I'm wrong, but wouldn't a filter sweep yield a more accurate (though still limited) impulse response?

It is a complete farce to claim that it duplicates the complex behavior of a modern synth filter from one single cycle waveform...

[hakey]

It is a complete farce to claim that it duplicates the complex behavior of a modern synth filter from one single cycle waveform...

To be fair, I don't think that anyone's claimed that. Tone2 only say that it's possible to mimic a filter.

[Frantz]

It is a complete farce to claim that it duplicates the complex behavior of a modern synth filter from one single cycle waveform...

Yes, Andy at Cytomic would not have spent years working on The Drop painstakingly modeling the behavior of the hardware components of each filter type if he could have just analyzed one single cycle waveform passed through each filter.

No doubt Rayblaster can produce interesting results with its "magical" process but to expect it to replace other synths and filters is an unrealistic expectation.

[highkoo]

Furthermore, I refuse to believe that it can allow us to communicate with the dead.
Or aliens, for that matter.

[Frantz]

To be fair, I don't think that anyone's claimed that. Tone2 only say that it's possible to mimic a filter.

Here's the claim on the Tone2 website:

IMS has no need for a separate filter section because its oscillators are capable of authentically reproducing the filter sound of other synthesizers by using one of the factory impulse presets or simply by importing one of your own impulses. Not only is RayBlaster capable of modeling the sound of existing filters - it also makes it possible to create completely new artificial 'fantasy' filters that are exclusive to RayBlaster. This renders its filter capabilities virtually unlimited.

[hakey]

To be fair, I don't think that anyone's claimed that. Tone2 only say that it's possible to mimic a filter.

Here's the claim on the Tone2 website:

IMS has no need for a separate filter section because its oscillators are capable of authentically reproducing the filter sound of other synthesizers

I stand corrected!

Naughty, naughty Tone2...

[MaxSynths]

Im not looking for a discussion, but consider that the Diva test took place because Diva is at the top of the heap, and for no other reason?

That test took place because Ingo, as a Jupiter owner, first created a Jupiter patch on Diva and now on Rayblaster. In short: that test took place because the Jupiter is "at the top of the heap" and Ingo is simpling trying to emulate that kind of sound with different virtual synths.