Does anyone else work with samples in “stacks”

Does anyone else work with single audio files for similar samples (like a bunch of boom bap kicks) where it contains multiple sounds played back to back that you can chop up in the MPC to different pads when you make programs. There are some advantages to doing this rather than working with lots of individual samples.

For instance, I’ll take a bunch of kicks and load them up in audacity and line them up consecutively then export them as 1 audio file called “boom bap kicks”. Then on the MPC to make a program only requires loading the 1 sample for tons of sounds. Then using transient slicing the file can be auto-sliced and you can choose which slice you want per pad as you build the kit.

It seems to make loading and saving faster, plus I can choose the appropriate “stack” for the style of drum I want. Generally I can have hundreds of drum sounds at my disposal in a project without needing to load new sounds using only a few audio files which use like 10mb.

I used to do this on my Korg ESX because of its file number limitations (which the MPC doesn’t really have).

Are there drawbacks to working like this versus having tons of individual samples?

The MPC only has 2 GB of RAM where all your samples are loaded to. You‘ll max out quicker, but if that‘s no bother for you so far, go for it, as long your storage is large enough.

Also, (destructive) processing might take longer or be a bit imprecise (e.g. normalizing, warp), but hey…

I can see the advantages you mentioned, but I personally tend to keep my projects slender. I usually don‘t resample multi-layered sounds (e.g. kicks), but keep my selected levellings and filters „at hand“ if I find it causes problems during mixing.

Interesting idea. I guess this is ultimately the same idea as chopping a break and using the slice numbers in a program to create a kit, which has the downside of being a bit more difficult to manage when assigning to pad layers as you are working with slice numbers rather than descriptive sample names, plus you have some limitations over what sample processes you can apply to slices in sample edit.

The advantage when slicing a break is that you can revisit the slices to tweak them, add more slices, use pad parameters or even use the entire break, all from that single break sample in memory. The 'stacks' idea doesn't have any of those creative benefits as you are just using a sample containing a bunch of kicks spliced together. So the start/end points are effectively fixed and there's no real creative opportunities to be had from re-visiting and re-slicing the 'master' file.

when I wrote my memory article I found that the MPC would reach a point where it refused to load, say, a single 100MB file, but would happily load two 50MB files, so there does seem to be some kind of buffer or overhead in there that means loading many smaller files is more memory efficient than loading a single file, however this may be purely a limitation that occurs during the initial loading process.

But I would imagine loading a single 10MB file is likely to be quicker than loading ten 1MB files. In web design we often would combine all our icons and logos into a single image file and then use background image positioning to display the relevant portion on the page - i.e. we only needed to load one image into the browser instead of 50, and the browser would render the page much more quickly. Stacks probably have a similar benefit.

I like the idea from a marketing perspective, selling 'stacks' containing related sounds, but I'm not convinced it offers any real benefits over loading a bunch of individual kicks. I think it's more a case of 'if it works better for you then go for it'.

I use chops a lot, i think it's one of Mpc features I like the most.
I sometimes use long samples (field recordings, clips, whatever) and make a new program with those chops,play them as an instrument, put different fx on different slices, reverse one, and so on.
As you said, many possibilities with just a sample on Ram, very funny and economic for ram.
Also I discovered that chopping and adjusting slices sometimes works better then warp, if you have to adapt a rythm or a clip to a different bpm

Thanks everyone.

Like MPC-Tutor suggested, I typically have a stack of kick sounds that have already been normalized and don’t need processing on the “master audio file” , or at least processing that can’t be done in the MPC in standalone.

Most of my kick stacks contain 20-40 kicks and are mono, sometimes even pitched up a 7th or an octave so they are effectively around 22k-32k when played back at pitch. Those files are typically 1-2mb. I’ll load a kick stack, a snare stack, a hats stack and a perc stack, so it may be as little as 4mb for all my drums.

The reason I ask about this approach is that I also apply this to key group instruments and multi-sampled drums. My boom bap and other drums are tiny in size compared to keygroup instruments. When I have a keygroup or multisampled drum set that loads and saves 40+ samples every time I save a project, it just seems to take a while, plus file management after is cluttered. I never edit the audio of any of my multisampled Keygroups or drum sets. The programs are one file and the accompanying audio is one file. Simple. Logic Pro does it like this for many of their multisampled drum sets.

Just wondered if anyone did this or if I’m heading down a path with potential pitfalls for a payoff that seems worth it to me.

Thanks.

I think that approach would definitely clean up the sample browsing window, especially with keygroups. I'm finding myself scrolling and scrolling that list to get through large keygroup programs, just to find a specific sample. Your approach would eliminate that at least.

One thing to note... if you do use a single .wav file to represent a key group or drum kit... it is possible you will get weird (though maybe it would be an effect) if you automated start/loop/end points as you could potentially scrub the entire length of the file... not an issue if you know about it and build for it... but just something to keep in mind.

T-Monk wrote: The programs are one file and the accompanying audio is one file. Simple. Logic Pro does it like this for many of their multisampled drum sets.

So these multisampled drum kits in Logic, are all the instrument articulations and round robins for one kit just stored in a single WAV file? So that one wav file could contain say hundreds of kicks, snare, hats, cymbals, toms etc? Or is there one wav file for kick round robins, another for snare round robins, etc?

MPC-Tutor wrote:So these multisampled drum kits in Logic, are all the instrument articulations and round robins for one kit just stored in a single WAV file? So that one wav file could contain say hundreds of kicks, snare, hats, cymbals, toms etc? Or is there one wav file for kick round robins, another for snare round robins, etc?

Yes, Logic puts all articulations and all sounds in a single file. It's a total mess the way they do it (perhaps to stop people from reverse engineering them.) They put kick 1v1, snare 13v12, hat8v15, etc. contiguously in one file where it isn't ascending velocities of the same sound. They also use .caf files FYI.

I don't put them all in one file, but I have created audio files with 16 sounds (kicks) at ascending velocities for a multisample kit. Then I assign the kick to Pad 1 of bank A, B, C and D. Playing Bank A Pad 1 simultaneously plays Bank B Pad 1, Bank C Pad 1 and Bank D Pad 1. For all of the pads I assign the one sample and per pad I select ascending slices for each of the 4 layers which I assign to a velocity range to correspond with the velocity I want to trigger the sample. (Bank 1 Pad 1 is slice 1,2,3,4; Band 2 Pad 1 is slice 5,6,7,8; etc. And Bank 1 Pad 1 4 layers are assigned velocity 1-8, 9-16, 17-25, etc.)

It takes about 15 minutes to setup a program (I created a template), but once it is done it is done and I have a multi-sampled drum kit with only 16 audio files, and a .pgm file. Much easier to deal with than 256 audio files and a .pgm file.

This has got me thinking. In PROGRAM EDIT > SAMPLES we have the 'SLICE' parameter on each layer, which currently can be set to Pad, All or a specific slice number. What if Akai added a fourth option: 'Random'?

So, if we had a 'stack' of 'hard velocity' snare round robins we assign that to LAYER 1 with Slice to set 'Random'. Set LAYER 1 to only trigger for velocities 100-127.

Repeat for the other three layers, e.g. layer 2 set to velocity range 75-100, configured to play a random slice from the 'medium hard' round robin snare stack.

This would finally give us a way to combine velocity switching and round robins. It would actually be very powerful, as each stack can have up to 128 round robins. Plus we can use simultaneous play to provide more velocity layers if we need them, or in the case of keygroup programs, assign multiple keygroups to the same note range and spread velocity ranges across them.

I've submitted a feature request for this - if anyone else thinks this is a good idea, please let Akai know.

Slice switching sounds like a better idea for sample alternation (round robin etc.) than layer switching, because there are only four layers, and that's not much. And four is a small power of two. I can easily think of interesting things to try with slice sets where the slice number could be "modulated", either randomly or according to a CC curve or something. Add something like, order the slices according to RMS power, detected pitch or something. Even wavetable synthesis comes to mind, although that needs some more thinking about cross-fading, loops and phase alignment.

I actually wondered what would happen with the current infrastructure if you got a sample with 128 hits at ascending velocities/volumes and set the velocity to effect sample start time. Would each ascending velocity 0-127 play each 128th of the sample? Have not tried it since it would require 128 different hits and not a lot of drum libraries have that many (excluding round robin samples) different velocity samples.

What would be interesting is to be able to trig sample start per event in the step sequencer directly and with the LFO

MPC-Tutor wrote:This has got me thinking. In PROGRAM EDIT > SAMPLES we have the 'SLICE' parameter on each layer, which currently can be set to Pad, All or a specific slice number. What if Akai added a fourth option: 'Random'?

So, if we had a 'stack' of 'hard velocity' snare round robins we assign that to LAYER 1 with Slice to set 'Random'. Set LAYER 1 to only trigger for velocities 100-127.

Repeat for the other three layers, e.g. layer 2 set to velocity range 75-100, configured to play a random slice from the 'medium hard' round robin snare stack.

This would finally give us a way to combine velocity switching and round robins. It would actually be very powerful, as each stack can have up to 128 round robins. Plus we can use simultaneous play to provide more velocity layers if we need them, or in the case of keygroup programs, assign multiple keygroups to the same note range and spread velocity ranges across them.

I've submitted a feature request for this - if anyone else thinks this is a good idea, please let Akai know.

very interesting idea! i hope akai reads this!