How can I sample my Virus TI?

[erol]

Hi,

I would like to make a Virus TI sample bank for Kontakt but I can't figure out how...

How can I (auto) sample my Virus TI?
Thanks

[Timo]

There's no auto way. However the 'when and what' to sample has become less of an issue now we have almost infinite software storage to play with. In less recent times you had to be a lot more brutal and decisive. A few years ago I carried out some multisampling from my Indigo to be used on my Korg Trinity which had only 8mb to play with in total, so both sample (keyzone) layout planning and sampling technique were as important as each other.

These days with software samplers you can use as much sample space as you like, but there are still techniques which remain common with the 'older' days.

Depending on the complexity or simplicity of the sound you want to capture, you have the option of sampling every single note (in otherwords 12 sampled notes per octave) through to 1 note every 12 (1 sample per octave) respectively. The former being for very complex sounds (12 samples per octave, usually reserved for real pianos, not synths, really), and the latter (one sample per octave) for simple raw oscillators or bass.

So, basically you can stick to sampling every semitone note (12 samples per octave) for really, really complex sounds, or every 2nd semitone (6 samples per octave), or 3rd note (4 samples per octave), or 4th (3 per octave), or 6th (2 per octave), or 12th (1 per octave) for less complex sounds.

So if sampling every 3rd note: C3, D#3, F#3, A3, C4, D#4, F#4, A4, C5, D#5... etc.

Or every 4th note: C3, E3, G#3, C4, E4, G#4, C5... and so on.

Using every 3rd note is usually good, as when keymapped it will pitch the sample up or down equally by one note on the keyboard. So a sample at C3, for example, would be pitched down to B2 and up to C#3. And the sample D#3 would follow on from this, being pitched down to D3 and up to F3, and so on.

Sampling C2 through to C7 is usually ok. Depends on both the complexity and frequency content of the sound.

The lower you go down the keyboard, say below C3, or moreso below C2, you don't necessarily need as many samples as those notes and frequencies are not so audibly critical so you can thin out the number of samples somewhat. If the sound is really complex, or has lots of top-end frequencies, then you may be wise to extend sampling the upper key range to avoid aliasing and chipmunking.

But the more notes you sample, the more memory you'll need to store the samples. It's a tradeoff. As mentioned it was harder in former times when having to create sample sets with just 8mb of available sample memory!

If you add velocity switching (multiple samples of each note at different velocities) then the memory needed increases exponentially.

More on what to sample....

It depends how much of the synth's flavour you wish to retain. The more flavour, the greater the complexity of sampling technique required to capture it in order to retain a usable sound. For example, you can sample just the sustain portion of the raw oscillators of a synth, with everything else 'wide open' (filters wide open, no modulation, no EGs, no LFOs, no effects, etc.) which is really easy and cheap in terms of sampling and memory required, and you can easily fake the amplifier attack, decay and release using the sampler's engine, but then you're also reliant on the sampler's own filters, effects and so on to process the sound when you play the sample(s) back, and you lose some or much of the flavour of the synth's original characteristics - like the filter and effects, as mentioned, which are sometimes as often as important (or even more important) than the oscillators themselves. This is as true with the Virus as it would be with a Moog or similar.

If you want to start sampling the synth's filter, it gets more complex, and you need to start making brutal decisions as to what to sample: EG settings, cutoff frequencies, resonance, filter velocities (whether you want to attempt to map velocity switches, requiring multiple samples per note), or include keyfollow, etc.

But whatever you can fake via programming, go that route instead - a lot of things you can fake easily as they don't impart significant amounts of flavour, such as the amplifier release, EGs, various LFO modulations, amp velocity, panning, etc. - so you don't necessarily need to sample those aspects, which means you can and should concentrate on sampling just the particular flavoursome parts, and then just program the sampler to carry out the basic, non-critical modulations.

If you wish to retain all the flavour, including filters, effects and/or arpeggiating, it then becomes much harder than you think - the number one reason being that everything on the Virus is free running, the oscillators, LFOs, time-variant modulations, effects, tempo and so forth - on most patches, every time you hit a new note on the Virus it sounds different. Sampling a note, and then playing this sample back using a keyboard will sound really mechanical as the phases of each element that went into the original sound will no longer be free running, they'll be fixed and will sound the same every time you trigger the sample. For example if you duplicate a sample to use as a 2nd waveform and detune it, the note will make a "peoww, peowww, peowww" sound each time you press a key, as the sample's phase is triggered at the same point each time before detuning away from each other. There is a trick you can do to hide this a tiny amount, which is to pitch the 2nd waveform up by 12 semitones (so it triggers a different sample from a different keyzone) and transpose it down 12 semitones back to the original pitch, which effectively triggers a different sample but at the same pitch, but it really is a crappy poor substitute.

Other flavoursome aspects are really hard to capture. For example there's a really nice sound on the Virus called 'TheDome HS' (as used by Rank 1 for Sensation 2003 anthem, I digress) - it uses an arpeggiator and the phaser effect - the fact that it's arpeggiated isn't really a problem if you have a programmable arpeggiator on your sampler, you can simply turn off the arpeggiator on the Virus and sample each note and then re-create the same arpeggiator pattern on the sampler. But try capturing the Virus' phaser effect... and it's nigh on impossible due to its free-running nature, yet it's such a critical component of the sound. Unless your sampler has an equally nice sounding phaser, you can't imitate something like that so well using samples at all. It wont sound like a Virus.

Again with 'Impact MS', which uses one of the Virus' own quirky LFOs, impossible to recreate without both the unique LFO waveshape and the Virus' classic filters.

Then when you bring stereo [unison/fx] into the equation etc.. Bah.

I could go on, looping at zero crossings, truncating, using crossfading, eliminating DC-offsets, etc. It's a massive subject, and requires a lot of hard work and time - preparing, sampling and reconstructing.

Basically, sampling and faithfully recreating the Virus effectively using samples, without resorting to one-trick samples, is no easy feat, unless you're sampling stabs, bass, innocuous pads and so on. What it comes down to, is basically if you want the Virus sound you have to get a Virus!

[nutrinoland]

Great Post Timo !

[LFO8]

There IS an autoway with Sample Robot or Redmatica's Autosampler

[Timo]

Yeah, those are more for sampling ROMplers. They wont give an effective result for sampling free-running synths - it would make said samples one-trick playback, giving them limited use rather than a sample-set to go back to and keep using - and furthermore the sample sizes would be completely non-optimised and humongous.

There are much better tricks for sampling ROMplers, though! Really crafty ones. But that's for another discussion...

After prepping Virus sounds, though, you can use automation to an extent - using MIDI to trigger notes (at 3rds or 4ths spacings) and record them all into a wave editor. Then the work starts.

[nutrinoland]

How about sampling the oscillator and then modulating it's phase with an LFO to mimic a free running oscillator ?

[Timo]

Originally Posted by nutrinoland

How about sampling the oscillator and then modulating it's phase with an LFO to mimic a free running oscillator ?

If that option is available on the sampler, that would be absolutely fantastic. Does Kontakt have that?

Hardware synths (like the Korg workstations), at least, don't tend to have that option. Samples on hardware tend to have a fixed pre-sustain region (for attack) followed by a loop region for indefinite sustain and release (programmed using an amp EG).

However, while modulating the phase (starting position) of a sample would be good for pads, strings and so forth that don't need to have a specific attack region of the sample, it might be disadvantageous for samples where the attack region of a sound is important, as it would randomly abruptly chop it off.

Thought it would be nice if you could allocate several slightly different samples of the same sound (with different phases) to a particular key, and then have it randomly select one of them when you hit that key. Or increment between them each time you hit the key. So they sound slightly different every time you hit the key. Of course the overall sample-set size would be a lot larger depending on how many multiple samples you allocate to each key, but it would sound more organic.

[MBTC]

Originally Posted by nutrinoland

How about sampling the oscillator and then modulating it's phase with an LFO to mimic a free running oscillator ?

I could be wrong, but at first glance I don't think this would have the desired effect, because a waveform of a sine wave is not really the same as a sine wave or any other natural signal from an oscillator, it has a designated start and end point which have data values between them (well data that represents data really).

For example look at the image on the web page below where it shows the basic wave types. Raw audio signals like a sine wave can be mathematically represented within the ranges of 1 to -1, so that the absolute value of data points between each zero point is the same (or the real values are symmetrical we could say, for example when the 0 to 1 portion is at .5 the 0 to -1 portion is at -.5, etc.)

http://en.wikipedia.org/wiki/Sawtoot...:Waveforms.svg

To the best of my understanding, all that flies out the window when dealing with any sort of audio clip. It might end up symmetrical enough to achieve this sort of phasing effect by matter of coincidence, but I don't think it would behave like an oscillator with most sound bytes.

That is, unless the data is read and synthesized.

I believe the plug-in Harmor does this (by image line). You can import a sample and it will basically read the data within and convert it into oscillator settings, then you can tweak the osc.

I think the best reason to use samples is because they are lighter on processing. The CPU doesn't have to "think" much (i.e. constantly compute the sine wave).

[nutrinoland]

On many synths, single cycle waves can be loaded as oscillators. 3osc, toxic , sytrus, surge, z3ta , serum etc.. I think it loops the wave. They generally have options for free running oscillators, retrigger or phase position.
I too am not really sure about what exactly is going on under the hood though. Harmor does single cycle waves and also does resynthesis, where it breaks down an audio sample like a drum hit , into single cycle waves, which You can then manipulate in different ways, including treat those waves like You would on an additive synth, modifying individual harmonics etc...You can also change playback speed , direction etc.. Kinda like granular synthesis , but not exactly. Morphine also does resynthesis, but not as well as harmor.
Maybe somebody else here knows about the difference between these types of oscillators.

[Timo]

I thought modulating the phase of a sample would be simply changing where the sample starts playing when you hit a key. A start offset. A start offset value of 180º, for example, would start the sample halfway when triggered - as long as the sample had loop data (otherwise the samples would end at different times each time you hit a key). Of course if you had samples with specific attack regions before loop regions (such as a piano sample) it would be fairly useless, but for strings, pads, synth leads et al it would be useful.

If you otherwise make a duplicate of a sample, and detune it from the original, and then trigger them both at the same time, you otherwise end up with that "peoww, peoww" effect when you hit keys, due to them both starting in phase and being summed together (causing a peak) before detuning away from each other. Not a great asset for ROMplers.

Modulating the sample start point would lessen this effect, but admittedly not get rid of it entirely as it's still a game of chance as to what phases (start offset) both samples are when you hit a key (if start points are modulated by LFO), although the longer the sample loop the less chance of the latter occurring.

That said, the Hypersaw works this way.

Whereas standard Unison (at least on the Virus) modulates the start offset (of an oscillator) in a very controlled way, almost like a triangle LFO waveform would, the Hypersaw's individual saws' start offset(s) on the other hand are all random. You can hit a key and find nearly all 9 oscillators are triggered in phase, but you can hit the note again straight afterwards and find they're all completely mixed (all have different phases) regardless of detune amount.

Unison and Hypersaw are two very different beasts. The latter would be much harder to sample.