Unreliable Sounds on Each Render

  • Basically every time I render a track my Virus sounds "change" ever so slightly. You wouldn't pick it up unless you were really paying attention, but if you are, it will be audible.


    Sometimes there's a harder twist or turn here and there, sometimes a bigger or smaller buzz on a buzzy lead. Other times, the sounds are blatantly different every time like a kaleidoscope. They have different sounds even though my automation curves are well-defined.


    I can't render my track 10 times in the hope that all the sounds will be exactly as I want them. My only other choice is to sample a certain sound and always use that sample, but that's tedious and you can't do it for every sound. The whole point is to have the TI open in your DAW and give a reliable result every time.


    Why are renders with the TI so unreliable (although on a very small, granular, almost imperceptible level) on some patches?

  • The Virus sports free running, fine tunable oscillators, free-running polyphonic unison, free running polyphonic LFOs, free-running delay lines and time-variant modulated chorus/phaser effect in stereo, stereo modulation, and a whole host of other programmed 'instabilities' put in place (via mod matrix or otherwise) to make the sounds less ordered and more organic. Natural oscillator phasing will also directly affect other audible aspects such as RingMod, etc. Even LFOs running in poly mode is enough to dramatically change things every time due to phasing, let alone anything else.


    Of course the Virus will sound different every time you play the same sequence, if you're paying detailed attention.


    You can reduce the free running modulation capabilities to an extent by asserting control over phase-inits of oscillators and LFOs, altering the polyphonic LFOs, Unison, and turning off analog tuning, and generally making everything sound more boring, but there are still things you cannot control, such as the phase cycles of chorus, phaser and delay and these will continue to modulate the sound in a free nature if you employ their usage.

  • Maybe I'm missing something but everything you mentioned is a fixed variable. Fixed variables, when plugged into the algorithm, are always supposed to give the same result. There's no randomness unless I specifically introduce a random variable and I'm not. So the same sequence should be identical.


    I'll look into the Phase-Init but that will take care of the obvious phase changes. I'm also talking about subtle, very grainy changes that are driving me crazy and forcing me to re-record 10 times, where there's no obvious phase culprit.


    At least with Nexus, which is a sampler, everything is 100% reliable from render to render.

  • That's because samples are fixed variables Once you've sampled something you cannot (largely) change it.


    Virtual analog synths, while they may have fixed settings, their individual free running nature, oscillator phasing and modulation will make it sound different each time.


    For example, think of Oscillator 1 being one module, and Oscillator 2 being another separate module. They're not linked (unless you use sync), and they run freely and independently of each other. Now think of each and every LFO (per voice) being separate modules, then add free-running unison, and then take each of the individual effects as completely separate independent modules. All of these modules are working independently of one another. The time variable is not fixed, unless you bring things like phase-init into play.


    Also when tempered mode is selected, the tuning of the oscillators randomly change +/-5cents too each time you hit a key.


    Why are subtle changes driving you crazy? Can you give an audio example?

  • This used to catch me out too… then realized actually this 'variability' is a good thing… and there are so many ways to introduce it on a VA :)
    … thats the issue with samplers for me, they are too 'static', but still, if you introduce an un sync'd filter/lfo on your nexus you will get the same 'issue'


    of course, if you want 'repeatability' all these things can be sync e.g. with clocked sources, also check the mod matrix is not using anakey/random.


    Timo, what do you mean by 'tempered mode' ? … i generally use Anakey for this.


    @eugenbalt, if you post at audio example, and the patch (in .mid format), it would be easier to hear what you are describing, and see if its a syncing issue.

  • Timo, what do you mean by 'tempered mode' ? … i generally use Anakey for this.


    Hi, I'm referring to the PureTuning (aka micro-tuning) option. Tempered (aka equal tempered) is default, as opposed to Natural and Pure.


    However Tempered mode also adds in subtle oscillator instability by default. The pitch of any given key will sound slightly different, by around +/-5 cents, each time you hit it.

  • This doesn't sound like the issue of free-running oscillators causing this. Free-running oscs are great for keeping the chord/detuning interactions natural on repeated notes. Most times this is great, other times, as has been said - phase init is good to restart the waveform for more percussive and predictable results.


    However another issue is here too. When you play too many multitimbral sounds concurrently on a Virus TI, the sound quality suffers. I hear it often with lfo control rate resolution, envelope 3/4 control rate resolution. How much you hear this depends on what sounds are playing together. and maybe even at what stage the sound is at. Snappy envelopes, fast lfos, recursive modulation probably require momentary bouts of dsp power whereas the rest of the patch may not.


    Maybe if Access implemented some more fixed limits wrt polyphony as opposed to ragging the DSP to overload this wouldn't happen. However, I guess the quoted polyphony would drop in order of magnitude, which from a marketing perspective would be a no no. However, my ears don't lie to me, I can hear these inconsistencies in the sound produced in a multitimbral situation.


    One way you can see if this is your problem is to track down each part one at a time and mix, then compare this to the TI playing it all out "live" so to speak.

  • UPDATE: 08-22-2015


    User "Daksha" kindly provided the below input and I was able to resolve my situation. TI2 Darkstar is working as advertised! Phew!


    Best,


    Naveman00


    ----
    08-21-2015


    Hiya everyone,


    So, I found this thread and it immediately struck a chord (pun intended)... I recently purchased the Virus TI2 DarkStar and when I attempt to either "render" the midi, which in Cubase 8 is basically by "exporting" the ONE Virus track from MIDI to Audio and then having it automatically paste the audio into the current project (at least this is the only way that I have found to work in some fashion, so that I can use a SEPARATE instance of the TI2 - SO freaking tedious!).


    Anyways, Eugene, I am experiencing the EXACT same phenomenon. Perfectly edited automation that sounds AMAZING in MIDI and after rendering the audio has subtle but IMO very noticeable changes. Not a click issue due to latency or anything, or so I believe. I just CANNOT get the TI2 MIDI to replicate to the SAME sounding arpeggiator in this instance. I have attempted DOZENS of workarounds and so far the result are audio sounds that I consider to be a horrible reflection of the TI2.


    I have contacted Steinberg and I will need to contact Access Support in near term because of this unacceptable rendering issue. My gut is telling me that Cubase and the TI2 must be "fighting" one another, as someone else mentioned I think. Again, IMHO it must be an underlying feature in BOTH of the products that is "shorting out" the translation from MIDI to AUDIO.


    If anyone from Access reads this, can you provide some feedback for the rest of us?! #Access #TI2 #Darkstar


    Thank you,


    Naveman00
    08-20-2015


    Cubase 8 Pro 8.0.20 (most up-to-date version)
    WIN7 / x64
    i7 / 32GB RAM

  • -Create a Bus channell in Cubase
    -Select the Virus track and set its output in Cubase to the Bus track you created
    -Select the Bus track and activate recording.
    -Record !
    This is what I do & always works.

    -Windows 10 Pro 64bit DAW (Asus ROG Maximus XI Hero + i9-9900)

    -Steinberg Cubase 10 + Steinberg CC121 Controller
    -Steinberg MR816 CSX + Behringer ADA8200
    -iConnectivity MIO10
    -Virus TI2 P0lar
    -Waldorf Pulse (1)
    -Waldorf Blofeld

    -Yamaha AN1x
    -Yamaha Reface DX
    -Clavia Nord Modular G1 Expanded

    -Moog Minitaur
    -Maschine MK3

    -Elektron Octatrack MKII

    -Adam A7X
    https://soundcloud.com/daksha

  • You are welcome! I'm happy this resolved your issues!

    -Windows 10 Pro 64bit DAW (Asus ROG Maximus XI Hero + i9-9900)

    -Steinberg Cubase 10 + Steinberg CC121 Controller
    -Steinberg MR816 CSX + Behringer ADA8200
    -iConnectivity MIO10
    -Virus TI2 P0lar
    -Waldorf Pulse (1)
    -Waldorf Blofeld

    -Yamaha AN1x
    -Yamaha Reface DX
    -Clavia Nord Modular G1 Expanded

    -Moog Minitaur
    -Maschine MK3

    -Elektron Octatrack MKII

    -Adam A7X
    https://soundcloud.com/daksha

  • You can also use a similar trick on Ableton, using an audio channel.


    I normally set the output to sends only and turn all the send amounts down to set (only way I've found to nullify the main Virus output).


    Then I set up 3 or more audio channels and take their inputs from Virus. Set up the channels for recording and off you go.


    Only draw-back, is having to crop the recorded audio since there's often some latency.