8 Step Variable Sequencers
  • I found the 16 step sequencers I made earlier were overkill particularly on the iPad where space is at a premium. I cut them both down to 8 steps for situations where the extra steps aren't needed. The functionality is the same otherwise (no chaining obviously).
    Screen Shot 2017-09-30 at 1.51.33 PM.png
    1392 x 1059 - 327K
    8 step variable sequencers.audulus
  • Awesome! I have found 8's enough, especially when they are being used to modulate other sequencers. I am looking forward to trying these out.
  • I originally made the chain-able model so I could program in a longer melody. It works for that, but I suspect it will become obsolete when we have better MIDI capabilities. It would be great if we could attach a MIDI file to a node and trigger playback, sync, loop, step etc.
  • Yeah, that would almost be as good as a data node. Notes could easily be repurposed for other functionality. Obvious examples would be filter-cutoff automation or triggering drum modules, but esoteric possibilities could be even more useful when creatively applied.
  • So have been using these sequencers a little and made some modifications for how I ended up using them. I think we have different expectations about how the gate output works, but that's personal preference. The things you might want to adapt for yourself are exponential scaling for the "clk len" knob and adding a sum of all the steps so you know your total sequence length. Oh, and I chromatically quantized the "o" output on the melody sequencer with one node, so that it leaves open the possibility of using it without a complex quantizer and save on CPU for the overall patch.

    Loving these, especially how clearly the inside is laid out, it makes modding a breeze.

    8 step variable sequencers modified version .audulus
    2732 x 2048 - 635K
  • So glad you're finding the modules useful. I've gone back and forth on the gate. I had it inverted originally so that it was high and pulsed low at the note change but eventually switched it back to a pulse high at the start of each note. It's a bit of a toss up when it comes to a sequencer. I assume you would want a high to signal the beginning of a note, but when should it drop back down? Ideally the end of one note is the beginning of the next. I'm certainly open to suggestions on what you think would be useful. The max clock length is perhaps a bit long, but it allows for triplets and dotted 64th notes. I like the idea of changing the scaling on the step length and a display of the total number of steps. The quantization is another one of those grey areas for me. My original design was quantized and then I decided to remove it on the theory that the outputs might be used to control things other than a note. Since there's a 0-1 modulator output I think the quantizer makes sense. Maybe I should output both. (come up with an SVG graphic for that :) ) In any case, thanks for the additions, I think they are all definite improvements to the original design.
    BTW, not to be overly pedantic, but multiplying a value by itself, while a useful way to scale things, is actually quadratic not exponential. A quadratic is y = ax^2 whereas an exponential is y = a^x ( a being a constant).
    P.S. I loaded your patch and now my dogs think someone is at the door. lol
  • Oh yeah, being able to subdivide is awesome with the longer click length parameter. But I just had trouble dialing in the smaller counts that are what I often need. Also, it's not easy to sum all those step lengths in my head and sometimes you need all the steps to add up to a certain number in order to stay in sync with some other sequencer.
  • I found myself adding up the steps too, so that's a useful add. What are your thoughts on the gate behavior?
  • On the Metropolis, there are 4 gate patterns for each step (which can have between 0-8 steps)
    1) gate low
    2) gate high
    3) play first clock pulse
    4) play all clock pulses

    and with the modifications I to the drum sequencer I can use the two sequencers in tandem to create the complete pattern. The gate patter can closely match the note sequence or it can be a separate stream. So what I'm looking for with the note sequencer is just a trigger at the start of the step.

    Attached is a patch that should illustrate what I'm getting at. It has 4 pairs of note and gate sequencers and the note trigger gate goes to a randomized switch that can be gate high, first pulse, or all pulses. The patterns all have the same length though, so it should loop the notes but not the rhythm.


    How can I get a trigger gate on the first step in the sequence?
    Look Mom No Quantizers.audulus
  • @RobertSyrett, I've incorporated your suggestions regarding the scaling and sum of clock steps, and reworked the gate logic. Both sequencers now generate a gate the lasts from the leading edge of the first clock in the step until the leading of the second step. The regular sequencer also generates a gate that is high from the leading edge of the first clock until the trailing edge of the last clock in the step. I also added a gate that is high at the first step of the sequencer. I included both a quantized and non-quantized "o" output and reworked the scale and offset controls to make them a little more useful. Thanks again for all the suggestions, and please let me know if you find any bugs! I found two today. I also included some text-only versions for our Windows friends.

    BTW I tried exponential scaling vs. the quadratic scaling you used and the quadratic is definitely better, at least in this case.
    8 Step Variable Sequencers V2.audulus
    Screen Shot 2017-10-02 at 4.18.12 PM.png
    1593 x 780 - 212K
  • Wow! Super awesomeness!

    In my feature request I should have been more clear. The reason why I wanted a first step gate output was to chain the sequencers together. However I did not account for the excellent quality of your sequencers, which actually play the step they are on when they receive a clock signal after being dormant, rather than the next step (which is a neat trick by the way) as all the sequencers I have made do. So I guess what I should have asked for was a EOS (end of sequence) gate.

    Attached are two versions of the same patch, one with my trusty old citrus seq and one with the new 8 step variable seq.
    sequence chaining.audulus
    sequence chaining with STS 8v.audulus
  • @RobertSyrett I'm not experiencing any sound on the second patch (with sts 8v) on my iPad. Is this intentional?
  • @Experiment1 yes, I'm trying to do the same thing as in the first patch, but I need a little help.
  • Exactly when do you want the gate? Current gate goes high with the leading edge of the first clock on step one. I might be able to make one go high with the trail edge of the last clock of the last step.
  • Yes, the trail edge of the last clock step would make more sense for what I'm trying to pull off. A million thank yous for the help.
  • @RobertSyrett, here's the modified sequencer. I'm sorry to say that your patch still doesn't run. I think you've encountered the dreaded Audulus feedback loop issue! :( I hit this same kind of problem several times when I was developing the chaining sequencer. The logic looks so obvious but there is something in the feedback path that causes Audulus to go off the rails. If you notice, the clock is obviously reaching the sequencer, I even put a waveform node on just to be sure. The LED is pulsing, yet the sequencer doesn't advance. If you disconnect the line to in2 on multiplexer, the sequencer will run until you re-connect it. I've had to step back and re-engineer several times to get around this. If you want some ideas take a look at the internals of the chaining sequencer MK II
    sequence chaining with STS 8v - modified.audulus
    chaining sequencer Mk II.audulus
  • Well it's been a day full of patching in Audulus. I'm afraid I am nearing the point of eyes crossing and mind fuzzing. Thanks for the insight into the issue. I still think there is hope because the citrus sequencer can do this type of chaining. I wonder if it's because it loses a step... Did you see @Experiment01's sequencer? It also has variable steps. I think when I have more caffeine in me I will conduct a comparative analysis.
  • I was never able to definitively determine what causes the issue. I suspect it's something in my design that is causing the race condition. I know that Taylor has made every effort to prevent these by inserting feedback delays when required, but given the flexibility of Audulus I can't say I'm too surprised when they occur. If you come up with any fixes please let me know. I would love to understand this better.
  • Sorry if I'm to much back to the basics... but what about this draft?

    These counters are chaining and are displaying three different states at their right (a0, a1, a2).

    They also don't reset to early, only right before their next gate comes (useful if you have looong notes).

    You can build the rest around it, for example like I did on my Metropolis brother (though it doesn't chain because I use another counter).

    In this draft is also a logic for "eating the first gate" - useful if your don't want to advance your seq on the first gate after reset and play the first step.
    1451 x 936 - 167K
  • I will lay down the gauntlet right now and say whoever comes up with the simplest, lowest CPU solution to this problem will get a free copy of Audulus 4 when it comes out.
  • I tried to strip down the UI as much as possible.
    - 8 Stages
    - 0-8 steps in each stage
    - chainable (connect A to B and choose one master)
    - Single mode (activate when he should run alone)

    Some small notes:
    - Stays in sync.
    - First note after reset is played. Seems like it obvious, but many sequencers I tried out of the forum skipped the first note.
    - You always get the amount of steps you've dialed in.
    - # of steps are S&H'ed and changes are only applied on a new cycle or reset.

    I did not implement a gate-modifying logic, I'd guess there is/will be a uModular version of it? I could implement anything like this, but then I would suggest switching to the EX1_Tetropolis.
    You can easily get the stage-number out... but for now... I wanted this sequencer to be as easy as possible.

    Sequencer 8 stage_variable step_chainable.audulus
    Seq 8 stage.PNG
    1565 x 969 - 300K
  • Dude, this could totally be a uModular sequencer. And I think you are on the right track, this seems to work they way I expected. You might have earned yourself the first copy of Audulus 4.
  • Here is the uModular version I hashed up.
    uExperiment1 Chainable 8-step Test.audulus
    Screen Shot 2017-10-04 at 3.19.03 PM.png
    835 x 950 - 229K
  • Great work! I obviously over-complicated things with my approach. Your's is a much cleaner design. @RobertSyrett I like the uModular version as well
  • @stschoen Now that it's chainable, I think it could get away with even being 4 steps. But yeah, I really like the design inside. Very helpfully annotated, I'll have to work on that for my own modules.
  • Glad you like it.

    The lights look like a falling raindrop or a TV remote. :P
    The vertical ones look more stylish but I'm not quite shure if it would fit my used way. I'm so used to the horizontal 'left to right' timeline - but that's just my opinion.

    You could just delete 4 of each and it should work ootb.
    We also should route the stage # to the surface. With it you could easily divide the signal with a logic like stage<=4 on a switch. That would keep your sequencer in place and you would save some chains.<br />
  • @Experiment1. Totally agree about the vertical thing, it's my attempt to go by @biminiroad's guidelines for his uModularset. Normally I like a flat row of knobs. But it's nice as a change of pace :)