How to do Round Robin in the Rytm Mk2 using the Sound Pool

Although there are some bits online and in this forum about Round Robin, I struggle to find a clear and concise way to do this, and especially focused on the Rytm Mk2, so wanted to place some information here for others and also point to some useful sounds to help with everything. Also maybe there are other ways to do this that might be useful to add that I don’t know about or maybe I get something wrong.

What is Round Robin?

If you are not sure what Round Robin is, it’s basically an efficient way to maintain the movement found in sounds coming out of analog circuitry that typically gets lost if using a single sample creating what is know as a “machine gun effect”. There is a really good example of the differences between the two using TR808 sounds.

808 Snare
808 Clap
808 closed Hat

Sample Chains didn’t work for me

After experimenting and failing with sample chains using OctaChainer v1.3 which is a great tool, but less so for this, I went a different route using the Sound Pool and a random LFO, which worked really well and is much easier to set up.

Getting some samples

For sounds I used the LonDonsen_Moore_TR808_Kit as these are free, quality, mono and already are setup for Round Robin use. I had the change the extension to .rar to open on my Macbook so I have also put these samples on my Github and should open without the need to change any extensions. They do include some NI Battery files in there as well, which I left in for transparency, but we don’t need these.

Setup and LFO settings

If loading a whole TR808 so to use every drum track on the Rytm Mk2, I’d recommend using 10 sounds per track, which fills up 120 of the 128 sound pools slots available. The samples linked above do have 16 available per sound so it might be useful to create another folder of 16 for some sounds you like using most, as with another folder it gets easier to navigate into that folder and simply “select all” if 10 or “select all” if 16, instead of clicking through one by one. 16 sounds will sound slightly better than 10, but 10 is good enough if a whole drum machines worth is running.

Let’s assume you have selected to use 10 clap samples. Load in all 10 claps in sequence to the Sound Pool, then in the track you wish to use the clap, load the 6th clap in that sequence to that track. When you hit the pad you should be hearing the 6th clap.

In that same track within the LFO section put in the following settings:

SPD (Speed) to 0
MUL and FAD as default
DST to Sample Slot
WAV (Waveform) to Random
SPH as default
MOD(Trig Mode) to TRIG
DEP(Depth) to +4.99

The LFO is going to be trigger each time selecting a sound between the first clap in the sample slot and the last clap.

Now you just do this for the next sound, the next, and so on, each time being sure to load the 6th in sequence of that next sound.

That’s it.

Some interesting behaviours

I did notice a few things. The Random LFO sometimes fires the same output more than once in a row. It might go 1, 4, 6, 10, 2, 2, 2, 4, 7, 5 etc. I don’t know why Elektron didn’t ensure that if a number is being used then +1 and if the number being used is the highest then -1…or something like this, to ensure no duplications, but there it is.

Another thing I noticed, and I think this is actually correct, if you set the DEP to 5 then on occasion it will play the wrong sound. It’s very rare, but it happens. So I think the 4.99 ensure that rounding works correctly, after all the LFO DEP goes to 127.99 not 128. I could be wrong on this but 4.99 seems to solve it.

Lastly, in a slightly non Round Robin style but sounds cool, you can use different LFO shapes, like the RAMP at set slightly different numbers in relation to the 16steps, to create a machine gun style sound that changes in it’s own timing, which is really nice. It immediately reminded me of Autechre, so obviously I will use that at some point haha.

If I have missed anything or need to clarify anything then shout and I will edit the post.

Enjoy.

9 Likes

Wait. What?

  • evenly spaced grid tab
  • the sample chain must be 120 samples
  • 16 bit, mono, 48 kHz

Now when you use it on the AR, you have to make sure the trig length isn’t too long or that the amp envelope is closed quickly enough, because AR samples parameter are start+end, not start/length unfortunately.
But it’s doable :slight_smile:


A trick to fill your Sound Pool

Place your Sounds dedicated to Sound Pool in e.g. Bank P: when you start a project just load the whole bank to the Sound Pool :slight_smile:

Hey

Nice tip on filling the Sound Pool, will try that myself.

With OctaChainer, the issue I hit was mapping the LFO to the start marker. The LFO is 0-127.99 and the window for the sample is 0-120.

How did you overcome this?

Well, I totally forgot to deactivate the synthesized part, so I still had my attacks pretty neat. :sweat_smile: :man_facepalming:

The depth is not really a problem, you could set it to 119 and it would work.
What you don’t want is starting reading a sample not at the beginning of it, thus have some kid of quantization of the output of the LFO. In other words, you want an integer number, not a decimal to be taken by the SMPL/STRT parameter.
But I don’t think this parameter accepts decimal values, so in theory it’s possible that it quantizes the incoming value. Let me check with your claps.

1 Like

If you could check that would be amazing, as all the tests I did it always missed the starts.

Reducing the LFO depth just moves in the total range of selection for which to oscillate. The only way I could think to get around this was if I added a gap of silence based on ratio between the sample range and the LFO range, although trying this it didn’t work. I have no idea what increment each step of the Random LFO makes? I’m assuming it’s possibly in steps of 10 but only due to 12 samples loaded I can manually move the L-marker in steps of 10 and it perfectly matches the start of each sample in the chain, getting the LFO to do that however…

It picks a random value between -DEPTH and +DEPTH most likely. So you should set SMPL/STRT to 60 and have LFO DEPTH to 60 I guess.

I have tested with a DEPTH of 1, it’s obvious the decimal output of the LFO is not quantized in the entry of SMPL/STRT parameter. So, you’re right, my bad: you can’t just work with a LFO on sample start, which is quite a pity. (Sorry for having hijacked this thread a bit) :face_with_hand_over_mouth:

This machine needs a quantized LFO! And an “SMLPL/LEN” parameter instead of “SMPL/END” :slight_smile:

2 Likes

Random is random, having two consecutive values is rare but possible.

The idea with the random LFO is that it’s bipolar. So if you set it to 5, you’ll get the range -5 → +5
If you want to choose between the first 11 samples, set SMPL/STRT to 6, and DEPTH to 5.
It’s better to have an odd number so that you have the same number before and after.

Your trick works because the value is quantized when used for sample slots (for the obvious reason that there can’t be decimals for a slot).

1 Like

No not at all, I think it’s good to have this all in one place so to clarify any confusions and anything that might have been tried. Happy for you/anyone to chip in, I was hoping some might, so thanks.

It would be great to know what the Random LFO outputs as it feels something relative otherwise I would expect the occasional clusters of start points, but it seems reasonably equal. Still, at this point it’s probably more curiosity as the other method works ok.

Ah yeah of course, I was missing the zero in the bipolar, that makes sense and also means an additional sound, which is nice. That’s helpful thanks.

2 Likes

This is amazing actually!

But I’m not able to unzip the zip file on my mac? It’s an 8.3mb file and my computer says that is an “unsupported format”. I’ve never had problems with zip files before.

Can anyone confirm that the zip file is ok?

1 Like

Hey

I found if using the samples from the source there was mention of changing the extension to .rar to get it open. I had the same issue.

The ones on my Github i rezipped from memory so they should open ok.

Edit: I’ve updated the original post so hopefully that’s clearer. Cheers for mentioning that.

1 Like

Interesting topic, thanks. I was also trying to figure out if sample chains in combination with lfo’s on sample starting points would create a good round robin effect.

1 Like

Glad it’s useful. :black_heart:

Effing this, I use random LFO also on synth parameters to make them sound more natural and DUH two identical sounds after each other is what I often hear. This problem of random numbers is well known. It could even be a completely different function to generate the “randomness” - because this is music.

1 Like

Yeah it feels a bit sloppy of them.
I often get that feeling with the Rytm, it’s function first, sound good second. Maybe thats a little harsh. They have pushed the bar up a whole load higher thats for sure. Still, it is what it is, no huge biggy.