Analog synths and round robin

[Shabdahbriah]

It's for a sampler which doesn't have any of those features.

It's for various samplers; Reason NN-XT, Sforzando, Maize, Kontakt, EXS24, etc. I even make soundfonts! Since I'm an independent developer I need to cover a wider range of formats.

Ah.. Sorry. For some reason I thought this query was Kontakt specific.

[himalaya]

Just keep in mind, that if you want to loop those samples, you will have a hell of a job with all those round robin versions of the same note that have to be looped.

It may not be much of an issue with simpler sounds, but for example, a 3 oscillator VCO sound will drift over time making smooth looping a real chore. I've had big problems like this when sampling Moog Model D.
Caching the right moment in the sound when the phase of all 3 oscillators was alligned in a way that a crossfade was meaningful, took a long time. I like EMU X3 for this as it has some very useful features that help in finding smooth, natural sounding loops (it has loop compression and different crossfade processes).

For this reason, I found it much better to use the sampler's internal features to introduce subtle per-note modulation rather than worry about round robin.

The end result is using these samples in a mix, where nobody will hear whether the sample uses round robin or not.

[Aryaroman]

Yes, but I might as well code a new synth plugin while were at it. Funnily, I'm currently building a wavetable synth in Flowstone (Synthmaker).

Looking forward to that!

[SampleScience]

Just keep in mind, that if you want to loop those samples, you will have a hell of a job with all those round robin versions of the same note that have to be looped.

It may not be much of an issue with simpler sounds, but for example, a 3 oscillator VCO sound will drift over time making smooth looping a real chore. I've had big problems like this when sampling Moog Model D.
Caching the right moment in the sound when the phase of all 3 oscillators was alligned in a way that a crossfade was meaningful, took a long time. I like EMU X3 for this as it has some very useful features that help in finding smooth, natural sounding loops (it has loop compression and different crossfade processes).

For this reason, I found it much better to use the sampler's internal features to introduce subtle per-note modulation rather than worry about round robin.

The end result is using these samples in a mix, where nobody will hear whether the sample uses round robin or not.

Yeah, I know. Perfect loops are hard to get too, the automatic sampler like SampleRobot don't do this well. It has to be done manually. That's why I don't think I would do more than 4 RR. Otherwise it would be a real nightmare.

[ras.s]

A bit off-topic but but about loop points, I'd like to mention this one piece of software that I found recently that makes finding those loop points much easier. Endless WAV (for Windows) by Björn Bojahr, it can be found here: http://www.bjoernbojahr.de/index_windowssoftware.html Not automatic, but its zoom view is simply brilliant.

[himalaya]

Endless Wav presents a similar workflow as Extreme Sample Converter.

Finding the actual loop point which gives a click-free loop result is easy to do in pretty much every sampler. What is the problem, is how the cross-fade is applied in terms of the actual DSP process to even out, or to minimise the effect of amplitude, phase, stereo, and harmonic differences in the two regions being 'connected' to make the actual loop.

For example, the aforementioned Emu X3 has what's called 'Loop Compression', which in combination with two different crossfade modes allows to deal with problematic loops.

Another example, and perhaps the best auto-loop algorithm I have seen, is the Penrose Machine option in Redmatica's Autosampler (now available in Apple Mainstage 3).
This is very good. The only problem is, how Autosampler actually deals with the length and trimming process. Here, it is totally broken. I don't think the people who coded this have ever used it to sample anything of a specific length because the issues are glaringly obvious. Anyway, I digress.

Yet another system is employed in Steinberg's Wavelab with some synthesis loop options. I thought this would be the holy grail of seamless loop creation, especially with big, synthesised sounds, but unfortunately the workflow totally kills it for me. I found it unusable.

Kontakt, is not even a contender here, which is strange, because you would've thought that the industry standard sample playback instrument would lead the way with sample loop manipulation but it's behind even ancient samplers like EMU X3. This is even more astounding in context of the expertise that NI have in synthesis. I'm thinking it would be child's play to develop industry leading loop creation DSP precesses with their knowledge. Maybe one day....

I've yet to test Falcon in this regard and I can't remember how Halion and Sample Robot deal with it.

So, I'd love to see more options where developers focus on how the loop needs to be 'massaged' with DSP recessing to get better looped sounds, at least Redmatica really tried to solve it and the Penrose Machine algorithm shows that. So it can be done.

But that's a topic for another thread I suppose.

[DSmolken]

I suspect the sounds which need RRs the most are ones that don't need looping anyway - short plucks, drums etc. Some sounds might need both, but you could always do 4 RRs for the short unlooped sounds only.

[himalaya]

He is sampling analog sounds and here, sustained leads and pads or bass drones are a must. So, round-robin or not, looping will have to be done, unless the samples are very long, which will mean a lot of wasted MBs.

[DSmolken]

Right then. Pads and drones are the kinds of things you can probably skip RRs for, and do a random sample start point instead, especially if there are free-running LFOs. Leads might have short repeated notes, so there RRs might be worth it.

[SampleScience]

Finding the actual loop point which gives a click-free loop result is easy to do in pretty much every sampler. What is the problem, is how the cross-fade...

Indeed, cross-fading loops is easy, but for some type of sounds it's more authentic to find loop points without cross-fading. It will just sound better.

[himalaya]

It's weird how that works sometimes, where no-crossfade is needed.
In fact, there were occasions when I would find a perfect loop without a crossfade on the fist selection of the cue points, I kid you not. It only happens when editing a gazillion of files and the mind developes a certain intuitive knowing where to put the markers. The Loop Borg is here to get us. We have been loopified.