Making Sample Banks of Your Analog Synths

For those who record sample material from their analog/hardware synths: what’s your workflow/approach?

Over the years I’ve become accustomed to jamming with my analog synths/drum machines and producing pretty much all of my tracks in real-time, without tracking/printing/sampling my gear (maybe the occasional exception here and there). Lately, I’ve been interested in building sample banks of recordings of my analog synths, especially for drums/percussion.

Do you find it’s worth losing the flexibility of adjusting synth parameters of the sound because you can generally apply EQ/dynamics/other processing to make the changes you need, or apply basic multisampling techniques? Does anyone, for example, record a few measures of a kick drum they synthesized on a monosynth/eurorack setup and then slice it in Simpler or something to capture the slight variation from one hit to another?

I really love the idea of getting an Atlantis in a 4ms Pod and using it to create banks of drum sounds and FX or other one-shots, and am even thinking about selling my AR to offset the purchase since at this point I basically use it for only a small subset of things it can do, but my control freak self is afraid of losing real-time flexibility.

Any tips or experiences in the realm of recording/multisampling/organizing in this context?

its something I do and have done for a while. As you say its a trade off; once the sounds are sampled you’re pretty limited in terms of what you can do with FX and dynamics although thats not to say you can’t mangle your sounds.

I do a similar thing with multiple hits of a drum. Sample a few and use them in different places to change things up a bit. I’ve just got an MPC Live and have been chopping up samples and using bits here and there. Its got an auto sample feature whereby the Live will sequence a string of notes over MIDI and sample then chop the resulting string. Not used it yet tho.

For years I used Renoise for making full tracks and relied on recording passages and sequencing the passages to make the songs. I think the ‘modern’ team is clip sequencing? I don’t know? Its just a workflow that I sussed and stuck with. Sometimes I’d edit the sequences to extract just single notes or hits and then sequence them…It just depended what I was doing or trying to do!

Im currently in the process of creating banks of drum and perc sounds on different synths and then sampling them with the Live. I could just as easily record the sounds into the computer and edit them there but its more fun with the MPC and I’m using it to sequence anyway so at least I don’t need to export them out of the computer. The MPC saves them as .wav files so I can back them up and use them in other applications in the future.

When its comes to synth sounds I tend to have the MIDI sequence running in sync with the recorded parts. I tweak until I’m happy with the sound then I sample. I can always replace the sample if I want to later on but I fin=d I can get interesting results after the fact by reversing/changing start and end points/fading in or out etc. Plenty of happy accidents to be found doing that!

2 Likes

1010 Blackbox

3 Likes

I sample my Tempest with mpc one his sequencer is limited, also polyphony is only 6 voices. I caputure several hits to cycle through them, and the fx on the tempest are unusable, so i make the best out of it. Commiting is good, it makes you progress and appreciate your sounds. My beat need a lot of zip zap sounds, and the Tempest ia really good at that. Bass is also punchy. I have thought if i should resample through the Analog 4, when i have some standard samples.

1 Like

A multisampler with an auto-sampler function saves a tremendous amount of time:

  • Apple MainStage and Logic Pro
  • Virsyn AudioLayer
  • current Akai MPC/Force and the free MPC Beats

This standalone software is free, but I haven’t tried it yet:

  • Tom Cosm’s External Resampler link

Advanced Sample Format Converter/Librarian:

Nice that with modern samplers with lots of RAM, we no longer need to spend hours finding suitable sustaining loop-points, fighting with zero-crossings (stereo samples grrrrr), etc. Auto-sampling makes it too easy. Edit the keygroup later, add filters, modulation, and FX, adjust your envelopes, etc.

Usually at least 3 samples per octave, but it depends on the instrument and the sampler. Sometimes 1 sample mapped across C-2 - G8 sounds fine in a busy mix.

For drums, I either build kits with 16 1-shots, or mega-kits with 64+ drums which are usually made after chopping a loop.

No. Usually synths sound more dynamic compared to samples. But it’s a worthy compromise.

Yes. For my mega-kits. Some samplers support robin-robin playback, but I like to have all the samples under my fingers.

Patience, adequate planning and organizing is pretty important. Sometimes just naming samples can be time-consuming, but it helps later.

Plan a folder structure that makes sense to you. Mine, it’s by sample type, instrument, sound, ie:

  • Drum 1-Shots > BD, SD, HHC, HHO, TOM, PERC, METAL, FX folders
  • Key Looped Multi-samples >
  • Key Looped Single-samples >
  • Single-Cycle Waveforms >
  • Sound FX >
  • Tempoed Loops >

Okay, I’ll stop now. :slight_smile: Oh, don’t forget to Add BPM and Key to your sample names. Oh, and a little laptop with a good audio editor is very handy and also useful for sample library backups.

5 Likes

It’s funny, for many many years when working more in the realm of hip hop I worked only with samples and the happy-accident-focused approach was all I knew, and I loved how inspiring and nonlinear it was. I think the years I’ve spent lately focusing on dance music have led me to lose touch of that so it could be a great project to rekindle that kind of process in the context of my current stylistic tendencies.

I’ve actually considered getting a Polyend Tracker to challenge myself a bit in that regard. Whenever I read/hear people who have worked extensively or primarily with trackers talk about that workflow I’ve found the idea terrifying but I could use a shake-up to the way I usually do things.

Might be worth seeing if I can figure out a way to do auto-sampling like this but in Ableton Live, or seeing if someone else has. Also Live 11’s comping feature would probably be perfect for making nice layers of live-sampled material.

This is definitely huge. I’ve driven myself nuts with endless tweaking and second-guessing until far too much time has gone by, makes the production of a single track drag on for ages. Sometimes you just gotta pick a sound and stick with it and get to the damn music-making part!

2 Likes

Excellent tips! Love the detail, I’m all about a meticulous process.

I’ve looked into various auto-sampling tools out there but hadn’t heard of this one, I’ll check it out. Just now discovered this method in Ableton but it may be overly elaborate compared to just using a dedicated tool.

I’ve been thinking on real time vs sampling workflow too. Currently got 4 synths (2 of which are stereo), 2 ch of modular and a mic/spare channel on a mixer, being bussed to the 4 inputs of an Octatrack which takes care of all the sequencing and master fx/compressor.
With that amount of gear I have to either select which 4 inputs I’m going to use for a particular track or set and work real time (which means having to remember changes to this), or sample good sounds as they come up and do my final arrangement in the OT.

I’ve landed at working real time for a live set (pick the gear and stick with it, using 3-4 thru tracks), and sample and lock it down once I have sounds I like for producing a track (up to 7 flex/static/neighbour tracks in which case muted MIDI tracks preserve the sequence and synth presets in case I want to change something).

Outside of that drums come from Rytm acting standalone with input from OT MIDI track 8 if I want (since it’s equivalent audio track is a master rather than a synth).

This gives me the flexibility to pick a different live setup as I feel like it and set up an OT project specific to that. I would currently pick the gear on tracks 1-4 direct to OT inputs, and run the stereo synths in mono since most live PAs are mono. If I really wanted I could use tracks 5-8 to sample the ‘opposite’ synths from 1-4, giving me more sounds.
Then I can have a different project for making finished tracks, and put any sound from any of my gear into a sample based production system in the OT with the tracks arranged as below.

It has taken me quite a while to reach this point!

OT tracks (= MIDI tracks)
1 - Avalon Bassline
2 - Volca FM
3 - DSI Tetra (ST)
4 - Access Virus (ST)
5 - Modular 1
6 - Modular 2
7 - Mic/vocal/other sample
8 - Master track

1 Like

What I’m wondering, instead of which chromatic ranged people here sample; what are the note lengths you sample? Do you sample long sausages to just apply envelopes to later, or do you sample different lengths per note?

I’d say unless I specifically want a short decay sound I tend to sample longer notes and then reduce their length. I used to do that in Renoise by adding an ‘off’ below the note trigger. Brilliant for having different note lengths at different points (whilst using the same sample) rather than being either stuck with too short a sample or having to set up different envelopes for different lengths. For me its bits like this that make trackers so immediate and why I stuck with Renoise for so long.

2 Likes

Thanks for the link. His method looks similar to the old-school, ‘quick’ way to create keygroups on samplers. Sometimes that’s the best option. Let us know how Cosm’s External Resampler works out.

Sample size used to be crucial, when dealing with a few MB of sample RAM, but it’s not as much of a factor nowadays - with more RAM we can just record longer samples, and trim them later. But usually, the lower the note, the longer the sample. Not too many hardcore samplists (or sample houses) do it anymore, but setting sustaining loop-points within a sample was the most time-consuming process, but it would allow for much shorter samples, and save RAM.

I really like the idea of a tracker in hardware like the Polyend but for me its not got enough tracks. If you’re okay using a computer I’d suggest giving the demo of Renoise a whirl and see if you gel with the whole tracker workflow. The way I approached it was to see it as a huge sampler with FX etc and a cool sequencer built in. So I would use the sampler feature to record .wav’s, drag them into the pool and just trigger them with the tracker. Its not that different to how I used to work with hardware sequencers and Akai samplers (and still do to an extent).

There a learning curve like everything else but (as someone who was new to using computers to sequence audio at the time) I found it really quick to pick up. The layout is very similar to what I’ve seen of the Polyend so as long as you can make the connection between the samples in the pool and the data events on the roll you’ll be fine. It helps that the pool is in a side bar and you can see whats being triggered.

The track limit in the Polyend is an issue for me as I got used to using a individual track for each drum or instrument so a kit might span 7 or 8 tracks with different FX and EQ settings per track. Thats before I start adding instruments!

Ah yeah a demo of Renoise is a good idea, have always been curious to see how I’d gel with it.

If you want a baby walk through it to get you going just PM me. Don’t get me wrong, I’m no power user but I can get you started.

2 Likes

Awesome, I appreciate that! I’ll pm you for sure if I end up diving in.

1 Like

Ha! This is interesting. I just imagined long sustained shrieking high notes and can imagine why:)

I have sampled most of my instruments commercially for the past couple of decades and as a result, I’ve gotten the technique down to a routine that works for me. I still loop some sounds (choirs, strings, etc).

My basic approach is to sample every C, E, F#, and A of each octave of the sound’s optimal range.

I tend to sample dry unless effects are a critical part of the sound. If it’s a sustained patch, I sample for as long as it takes to cycle through a loopable section of the sound (letting LFO’s, envelopes, and other modulations play through.

I specifically design synth sounds to be sampled, considering the motion, texture, and how difficult in might be to loop the sound if it’s a sustained sound.

I then top and trim the silence off the sample, normalise, correct pitch issues, set my loop points add embedded pitch information and save the file with the note name and number in the file-name.

For tough loops, I use an old G3 iMac which has a (licensed) copy of Antares now-defunct Infinity, a great bit of software that could loop difficult sounds (like human voices). Incidentally, that same iMac has Turbosynth on it, too.

When mapping I might crossfade layers, but generally not. When using one same for multiple notes, I tend to map down from the root note to the next lowest sample. I find that this sounds more natural than trying to pitch a sample up from the root note.

I almost never use artificial time-stretching or pitch shifting to map a sample except for effect. I like to keep things as pure and natural as I can.

If it’s a sound with a lot of potential variations, I might assign multiple layers to a single key and use some modulation to select which is playing.

I very much like to build new sounds from my sampled synths in the target sampler. To me, the sampled sound is often just the start - fodder for the possibilities offered by the sampler in which it will be played.

For my own personal use, I most often sample my old synths as a compositional aid. For instance, if I know I am going to make an entire song using only a specific mono synth, I will sample patches I am likely to use and then work out the composition in a DAW or with a sampler/sequencer combo. Once I have the arrangement worked out, I’ll go back and re-build (and improve) the various patches, tracking them one-at-a-time.

I am so not proof-reading any of that… :stuck_out_tongue:

14 Likes

how come you chose these notes (c,e,f#,a)? this made me curious

1 Like

It seemed a reasonably even spread of notes with only a single sharp, which meant less typing. When you’re dealing with hundreds or thousands of samples in a given job, that actually adds a lot of time to the task. Plus, many samplers that do not recognise embedded pitch will default to “C” as the root note, which gave the users the ability to simply dump in a single sample (or only octaves) if they didn’t want to get bogged down in mapping for their particular sampler. So my logic was to start with the obvious © and work up (or down) from there. I went up to E because it was fairly commonly played. F# is halfway up from C (allowing the possibility of only sampling two notes per octave where appropriate while maintaining some common ground). A is roughly equidistant between F# and C, while still being a natural and, again, is commonly used. By ensuring that most commonly used roots are available, there is more chance that any music made with the sets would sound natural and unprocessed.

In other words, I probably ‘over-thought’ it. :blush:

7 Likes