Internal Audio Routing between parts

  • Hi,


    Each time I use the Virus as an audio FX Unit (routed from my mixer's send), I feel like it would be great to route a part's audio OUT to another one's audio IN, internaly. This way you could somehow 'chain' the filters or the distortions, and beef the whole thing up...


    Cheers

  • Hi Kilow,


    Thanks for the support.
    I'd just like to point out the difference between:
    - routing modularity inside a single part (delay before reverb, or frequency shifter before chorus),
    - routing modularity between parts (routing internaly part 1 outputs through part 2 inputs)


    Cheers


    MH

  • This option absolutely does exist on the versions before TI. Why Access chose to leave it off the TI is honestly incomprehensible to me. On the C you could, for example, route the output of the delay to two additional left and right parts, then cross-pan those, and feed them back to the delay unit, creating a cross-panned feedback loop. Add filter, ring mod, and distortion to taste. Madness.


    You absolutely cannot do this on the TI.

  • Why Access chose to leave it off the TI is honestly incomprehensible to me


    because sometimes you have to remove things in order to improve other things. in this case we found Total Integration and a voice count which more than doubled from the Virus C to Virus TI good reasons to omitting AUX sends. by the way: nothing holds you back from using the analog outs/ins to feed audio from one part into another etc. - it is not 100% the same but it should get you somewhere interesting.
    marc

  • Thanks for the reply Marc!
    I don't follow your logic - neither integration nor higher voice count let you do the sounds that you could do with the C but that you can't with the TI - unless there's something I'm missing? If Ferrari released a new car that had a better engine, but only gave the transmission two gears, some of their drivers might find that decision odd.


    I find this decision by access odd. Having an internal aux, even if most people don't use it, was an extremely powerful option. Consider how much more powerful the TI would be with it - you could, for example set up a multi with a pad sound being sent to an atomizer - and then set key splits so you could play the pad and the atomizer glitch at the same time. Or do the same with the vocoder going to the atomizer.


    It's a frustrating omission to me because it's a programming, not a hardware deficiency. There's no reason the option couldn't be re-enabled (It's the same series of DSP chips as the C, no?). Sure, using that option would cost some voices, just as it did on the C, but why not give your customers the choice?


  • in this case we found Total Integration and a voice count which more than doubled from the Virus C to Virus TI good reasons to omitting AUX sends.
    marc


    ... but in this case aux sends would actually help increase voice count...no? since having an fx (e.g. reverb) on 16 separate channels is much less efficient than having a single fx that N channels use.
    ( I think sometimes, this is a contributor for some getting low poly count, as they put multi fx on multiple channels, it really eats up dsp resources fast :))
    the drawback of using external loops, is you loose a virus output (or a pair if using stereo sends), and if you want to also have other (stereo) external input, you need a mixer.


    that said, I quite like using this approach with a mixer anyway, as it provides more flexibility to either route to other fx units, or to mix other inputs into the virus.
    it just a bit more 'hassle' to setup.

  • ... but in this case aux sends would actually help increase voice count...no? since having an fx (e.g. reverb) on 16 separate channels is much less efficient than having a single fx that N channels use.


    I think Marc was talking more from a marketing standpoint - I imagine that most people who use the virus as a stage synth or integrated with their computer never missed the auxes, either because they're too esoteric, or having the option to route audio in a daw made up for it, respectively.


    Using an input part reserves one voice for each channel (ie 2 is it's a stereo input) *all the time*, so you can eat up voices pretty quick with the routing.


    I've considered using a mixer with the TI, but even with, you don't quite get the same functionality because the auxes on the C let you route wet signal from the delay/reverb and dry signals separately - presumably to help with mixing down, but it also lets you process the effects in a feedback loop. For example, in the video below (sorry it may not be my best, it was early days), I have delay/reverb routed back through an input channel so the 'FX send' on that channel becomes additional feedback control, and lets me use the filters on that part to 'sweep' the feedback (best heard at the end of the video).



    The additions to the delays on the TI are nice, you can do similar things with the modeled tape delays by adjusting frequency and bandwidth (which is technically mislabeled, because increasing this actually reduces bandwidth of the filter, but whatevs), but if you try to sweep it, the feedback drops off unnaturally sharply. You have to move the filter control *very* slowly to get the same effect.


    For now, I think I'm resigned to keeping the TI2 around for a little while - I considered selling for an indigo2 and TI snow (I do really like the new oscillator models and the drive/distortion from the kemper amps are quite spectacular), but keeping 2 virus synths and all their compatible/non-compatible patches straight in my head might be more than I can bear. Definitely going to grab a Jomox T-resonator though, which should largely let me add back the delay sweeps I had before. Who knows, thought - watching that video and pulling the old bitstream out makes me super nostalgic for that old C magic :)

  • I vote for bringing internal aux'es back... or reimplementing it - whatever is easier to do. This is something I would gladly exchange a number of voices for.


    In all honesty I have zero hope this will ever happen - they clearly redid all the code for Multi Mode on the TI to make way for per-part effects, so it's far from a trivial addition. It would also mean significant changes to how multis are represented in the TI plugin. Would be cool though, it meant that multis could be programmed to be more than just a sum of their parts.

  • interesting @bctracks, not played with a C , so not seen the aux buses...


    >>Using an input part reserves one voice for each channel (ie 2 is it's a stereo input) *all the time*, so you can eat up voices pretty quick with the routing.


    ? don't get you on that, is this on the C ?
    on the TI with external input, there is no "voice" (excluding atomiser/envelope follower/vocoder fun)... the input patch is just using dsp resources, to process a stereo audio stream.
    ... the same as if it was doing it for a single part... the only overhead would be the additional ADC DAC (due to going in/out the analogue ports)


    with a mixer, you decide on the mixer how much wet/dry and feedback you want to happen, so thats a not an issue, (as you use say out1/2 as dry, and out 3/4 as wet)
    and of course you can mix in with other effects units, which is why I like it..
    BUT it sounds like you could have multiple fx buses on the C? in which case, yes this is a limitation, as we only have one stereo input to play with, so you can mix things in a different points in the chain.


    I do agree unlikely to happen though, and as you say the 'per part' is also not something I would want to loose, generally its easier to use,


    but would be nice if on a part rather than select OUT1/2,OUT3/4 we could select PART1-16, and then PART16 has this as input, as if it came from the external jacks.BUT as you say it probably is quite hard to implement, as its possible the TI only can do this with one input (stereo) channel.
    anyway, as you say .. not going to happen anyway ;)

  • interesting
      bctracks , not played with a C , so not seen the aux buses...


    ? don't get you on that, is this on the C ?
    on the TI with external input, there is no "voice"


    It's just a different paradigm on the C - I'm saying 'voice' because there
    is an extensive section in the manual on interpreting how many 'voices'
    of the synth's 32 activating each feature uses (osc3, analog LPF,
    etc...), it boils down to effectively the same thing - input parts
    aren't free, they use DSP too :) There wasn't an easy view of a patch's
    complexity, so it was important to keep this in mind when programming.


    Quote


    with a mixer, you decide on the mixer how much wet/dry and feedback you want
    to happen, so thats a not an issue, (as you use say out1/2 as dry, and
    out 3/4 as wet)


    It's true, the distinction between using a mixer to loop back and using the
    internal auxes is pretty small. With the auxes you get the dry output of
    the input channel without delay, but with the virus's filters,
    distortion, etc. With a mixer, you can get the dry output without any of
    the virus's effects (from the mixer). In the end, it requires another
    piece of equipment which should really not be necessary. I've sketched
    out the two feedback implementations for the C (top) and TI (bottom)
    below...


    http://www.brendanclarke.com/files/20150321_205327.jpg


    In an ideal world, the TI would have an output select for each part's
    delay/reverb block in addition to the part output but again, I doubt
    that's going to happen with all the changes that would mean to the
    summing structure

  • I think Marc was talking more from a marketing standpoint - I imagine that most people who use the virus as a stage synth or integrated with their computer never missed the auxes, either because they're too esoteric, or having the option to route audio in a daw made up for it, respectively.


    the Virus C has one DSP and the AUX sends more or less came "for free". the Virus TI has two DSPs and implementing AUX sends would mean that one has to send loads of data from one DSP to the other. my comment in regards for voices was an attempt to indicate that with opting for a dual DSP setup, the possibility for also having AUX sends went straight downhill.
    marc


  • the Virus C has one DSP and the AUX sends more or less came "for free". the Virus TI has two DSPs and implementing AUX sends would mean that one has to send loads of data from one DSP to the other. my comment in regards for voices was an attempt to indicate that with opting for a dual DSP setup, the possibility for also having AUX sends went straight downhill.
    marc


    Does TI distribute DSP load of odd parts to one DSP and of even parts to the second one irrespectively of the mode of work (with Virus Control or standalone)?
    If yes then could there be separate AUXes for odd and even parts? Though, this could be a bit awkward..

  • Yeah, I see, there's really no way to bring it back in the current model then. Maybe the snow, since it's got only 1 chip? :D


    It's a shame, I felt the C was more well-thought in it's implementation of multi-timbrality than any other synth. Because what you'd do if you had each timbre as an individual instrument would be to plug them all into a mixer, and with the C, you got a well-featured mixer, with auxiliary send and returns, etc... inside already!


    So... if anything comes of this, please don't skimp on the TI3! All the good stuff of the TI2 would be well-served with soild mixer implementation! And then, hey... you can sell a Kemper digital mixer out of the deal, too ;)


  • Does TI distribute DSP load of odd parts to one DSP and of even parts to the second one irrespectively of the mode of work (with Virus Control or standalone)?
    If yes then could there be separate AUXes for odd and even parts? Though, this could be a bit awkward..


    the odd/even split is the 'first stage' of allocation, but the TI can shift voices to the other DSP when load becomes high, so you cannot rely on a single part going to a particular DSP... so complicates implementation.


    i think if it was done (and it wont be), you'd have to move it onto one dsp, which implies a delay of one sample buffer*... its all possible, basically its the same process as is used when using a (most) DAWs which uses separate threads for each track and sends, but its not trivial.
    * of course if you route external out to in, you also get this delay.


    the other possibility, would be for the TI to expose the DSP allocation, and allow the user to lock parts to a particular DSP, and then insist a FX part can only be fed by parts locked to the same DSP. Sure, its an advanced mode, as auto-allocation generally is better BUT if you know what you want perhaps this is something you'd be prepared to sacrifice. (its not that complicated... kind of similar to how on a Spectralis/Elektron A4 you can do explicit OSC allocation)


    anyway, not going to happen, and my suspicion is if there ever is a TI3, it will have more DSPs (four?), so hopefully the DSP allocation would be reviewed then to allow for a little more control for advanced use.

  • Yeah, this is just what I was thinking - it would be great if the next version had a dedicated 'master section' DSP to handle routing and master effects. It would be a worthy addition to the series, rather than just slapping more voice power in there and calling it a day.


    There's a huge range of things they could do with a master effects section - multiband dynamics, a more natural home for the surround settings and atomizer, even open up the possibility of sequenced effects a la Roland's new mixer ... thing.


    But of course, I understand I'm firmly in fantasy land here.