Single Cycle waveforms

[sl23]

I have around 800 Roland Mc 909 samples, including various pitches from C1 ~ C5 mostly.

I decided to cut them into single cycles for use in MuLab's MuSynth so spent all of yesterday slicing then up, then afterwards remembering they need tuning. First time I've done this, never used sampler before. So looks like I need to start again!

How do I go about tuning these single cycles? I've Googled all morning trying to search for help or a guide but can't find any useful info. Of course, there are many sites with general advice but they all leave out specifics.

I remember someone here, possibly dakkra, mentioned about retuning adventure kids waveforms from D2 to C4, would you mind explaining the proper process please.

Also, is it worth using all the various pitched samples or just the C4 ones? Would there be much variation to require the others be worth editing? And... Some samples don't have a C4 sample, can these be frequency shifted somehow?

Some of my waves require two or more cycles to be more accurate which causes a difference in pitch from shorter waves? Which is correct?

I realise single cycles are great on simple waves, but when it's a patch with a varying waveform, how can it be reduced and tuned? I find it clicks/buzzes if too long and if it's a single cycle it doesn't replicate the original patch.

I can only assume that the single cycle format is limited in its possibilities?

Thanks for any advice

[dakkra]

So a true single cycle waveform plays back at the frequency assigned to the note event it processes at the moment. Pitch differences can happen when you don't use a true single cycle waveform. For example, If you put 1.5 Saw waves into a single cycle slot, the perceived frequency will be off as it plays 1.5 cycles at frequency F instead of 1 cycle.

Let assume you want to play a wave at 20hz

let c = the cycle sample, f = the frequency it's played at, and p = the perceived pitch

Thus: c * f = p;

So for a true single cycle instance, 1 * 20 = 20;

But if you have 1.5 cycles in your single cycle sample buffer: 1.5 * 20 = 30;

Hope this explains the concept well,

Dakkra

Edit: This also means that if you place 2 complete cycles into a single cycle slot, you will hear it at 2-times the frequency you play it at (eg: 2 * 20 = 40). This phenomena is also how hardsync is created, by offsetting real-cycles and single-cycles

[sl23]

So it doesn't matter how long a single cycle is? It obviously depends on the original wave. I get that part and the frequency part. Thanks for explaining. Some patches need two or more cycles though to replicate the original sound, that then reduces the frequency, which I imagine affects the tuning of the MuSynth presets,yes?

You mention that using longer samples increase frequency, but my experience seems the opposite?

Forgive me, my knowledge of synthesis is very poor. So please explain in layman's terms if you can.

Could you explain how to use this to reduce samples that appear as a random wave, as I have quite a few patches that aren't simply variations on square, sine and saw waves, but are seemingly random waves that perhaps use noise or LFO's to create the original patches.
Is it even possible to replicate these types of patches using the single cycle process?

Thanks for your help

[dakkra]

I tried to make a video but my recording software decided to crap out on me.

So for perfect, non-chancing waveforms, a single cycle will do.

If you need multiple cycles to get the sound (like if there's phasing or a cycle-offset) you would put each cycle into an array of wave tables. That way you can use modulation to morph between the different types of single cycle wave forms that are present. (This is also why wavetable/wave-morhping synthesiszers became popular last year, because it made multi-cycle synthesis easy).

So, using an MFO in the MUX, create an array of wave-tables that contain each individual single cycle waveform you need to include, then use modulation to change the wave-index while playing note events.

Hope this helps

Dakkra

[pljones]

You mention that using longer samples increase frequency, but my experience seems the opposite?

He said more actual cycles in the waveform. The frequency of 20Hz - 20 assumed cycles per second - remained. So if you have 2 actual cycles in a waveform that is assumed to contain 1 cycle and you play it 20 times in a second, you'll get 40Hz not 20Hz. However, if you have a waveform that only gets played 5 times in a second because it's actually longer, then the frequency does not follow the function given.

[BertKoor]

Shameless self-promotion: http://www.bertkoor.nl/MusicCalc.html
This enables you to calculate the frequencies & notes based on the length in samples of your single (or multiple) cycle waveform. That way you can instruct your sampler how much detune is needed, etc etc.

Instructions: fill in what you know, and the calculator will figure it all out by itself.

[sl23]

Thank you very much for your replies.

But, there's no mention of HOW to adjust the frequency, unless I'm missing something. This is all new stuff to me so finding this a puzzle! Is it a simple matter of using your oscillator parameters to adjust pitch?

Wavetables are something I've never used before either, so if there's any more info or links you could share that would be good. Think I got a lot to learn!

Thanks

[Taifunk]

Is it a simple matter of using your oscillator parameters to adjust pitch?

Yes

[BertKoor]

there's no mention of HOW to adjust the frequency

Well, what sampler are you using?

Usually, in the sampler you specify the WAV containing the sample. Then the sampler must know what the recorded pitch is of the sample. You have to tell it that. You can use the MusicCalc thingy I linked to previously to calculate what the pitch of such a single-cycle waveform is.

[Leslie Sanford]

Quick reply here, this may have already been mentioned, but the sample rate you use needs to be a multiple of the single cycle frequency.

For example, middle C is 261.626. If you use 44100 as the sample rate, dividing the sample rate by middle C gives you: 168.56. Thus middle C is not a multiple of the sample rate and can't be perfectly looped.

[sl23]

Thanks for all your advice, it's greatly appreciated.

there's no mention of HOW to adjust the frequency

Well, what sampler are you using?

I mentioned above that it's for MuLab's MuSynth. It allows you to use your own waveforms and saves them as part of the preset. My plan was to use all my samples and any future sampled patches in this Synth for ease of use. I'm obviously no expert with in music production, just a bit of fun, so I'm not too fussed about sound sources, quality and such, I just wanted to try and these patches right so I could share them with other MuLab/Mux users.

One more thing, is there much benefit in creating synth patches using samples from C1 to C5 or is it adequate to just make one patch using the C4 sample?

Assimilation nearly complete...

[sl23]

I've just found a plugin called GTune by GVST, using this in OcenAudio editor shows a C4 sample I edited to actually be playing at F# 94.83Hz. So to get this sample to the correct pitch it needs to be 440Hz, yes? Or am I completely wrong?

[BertKoor]

That would be quite right if you edited it to be 465 samples long.
But you don't really need it to be any pitch. It is the pitch it is, and the sampler should be able to play it at any pitch. You just need to tell it it's F# +43 cents

[sl23]

I see, so all this talk of tuning is pretty much redundant then?

But what confuses me is when you need multiple cycles, some samples only repeat after two or more cycles, it affects playback pitch. So if I had Osc1 play a single cycle of Sample1 it'd play back differently than the same sample but twice the length. So a two cycle sample plays at a higher pitch, but then this would affect actual playing pitch in a synth/sampler?

Perhaps I'm overthinking it?

My main issue here is how do you edit a random type sample down to a few cycles to use as the Osc wave in a synth? Something like orchestral ensemble for example. Is that possible?

Thanks for helping me with this

[dmbaer]

So a true single cycle waveform plays back at the frequency assigned to the note event it processes at the moment. Pitch differences can happen when you don't use a true single cycle waveform. For example, If you put 1.5 Saw waves into a single cycle slot, the perceived frequency will be off as it plays 1.5 cycles at frequency F instead of 1 cycle.

Sorry, but this is just plain wrong. Consider any multiple-oscillator synth where an oscillator can be synced as slave to another as master (a very common feature in both hardware and software synths). A slave oscillator restarts its playback whenever the master starts a new cycle. So, if we had a saw wave in the slave and it was tuned to play at 1.5 the speed of the master, we'd get one and a half saw waves in the slave, and it process would repeat.

As you adjust the tuning of the slave oscillator between more than one and less than two cycles per master oscillator cycle, the pitch does not change, but the timbre does. At exactly two cycles in the slave, the pitch produced by the slave oscillator goes up one octave. But in between one and two, the pitch remains that of the master. It does sound a good deal "edgier" however, which is what the oscillator sync feature is all about.

Extending the length of any single-cycle waveform, repeating a portion of the initial full cycle to create a new waveform results in a new and different single-cycle waveform. Single cycle waveforms always are heard at the pitch of the oscillator playing them (unless they contain multiple, repeated single cycles, of course, in which case it's not really a single-cycle waveform in the first place).