Show spoken script
Welcome back. This one’s advanced, and it’s going to feel a little strict on purpose.
Today we’re building percussion call and response from scratch for drum and bass in Ableton Live, but with one hard rule: resampling only. No hunting for percussion loops. No foley downloads. No “let me just grab a shaker real quick.” We’re going to generate an entire percussion conversation from the drums you already have in the project, and then reprint and mutate them until they become their own language.
If you do this right, your percussion stops sounding like random layers and starts sounding like a designed system. Like it belongs to the track.
Alright, set your tempo first. Put it in the classic rolling zone: 172 to 176 BPM. I’ll sit at 174.
Now, quick vibe choice: do you want it dead-straight and modern, or slightly swung and human? If you’ve got a groove template you love, cool, but don’t overdo it. If you’re using the Groove Pool, try a subtle Swing 16 groove and keep timing amount around 10 to 20 percent, random just a little, like 2 to 6 percent. This is one of those “seasoning” moves. You’re not trying to hear the groove template; you’re trying to feel it.
Next, we need a minimal source kit. Think of it like ink. We’re going to make a lot of drawings with a tiny bit of ink.
Create one MIDI track and name it DRUM SOURCE, Rack. Drop a Drum Rack on it.
We only need a few cells:
A tight kick, not boomy.
A snare with a proper DnB crack.
A closed hat.
A rim or clave, or even just a short click.
And one “noise” hit. If you don’t have a noise sample, cheat it from what you’ve got: take a hat, pitch it down, filter it, and now it’s a noise stab.
Do a tiny bit of shaping, but keep it light because we want the resampling stages to do the heavy lifting.
On individual cells, a Saturator with one to three dB of drive, soft clip on, is plenty.
EQ Eight on hats and rims: high-pass them around 200 to 400 Hz so you don’t build low-mid fog later. If your hats need a little shine, a gentle presence bump around 7 to 10k can help, but again: light touch.
Now we write the Call. One bar. This is the identity. This part matters more than people think, because if the call isn’t recognizable, the response won’t read as a response. It’ll just sound like… more percussion.
So before you make it cool, make it readable.
Create a 1-bar MIDI clip on your DRUM SOURCE rack. We’re focusing on hats, rim, and ghost energy around the snare. Your main snare for the full drum groove is still on 2 and 4, but for this percussion layer, we’re going to dance around that pocket, not stomp on it.
Here’s a solid starting idea at a 1/16 grid.
Put your closed hat on steady 16ths, but remove a couple hits so it breathes. For example, remove the hat on 1.3.3 and 1.4.3. Those little gaps create motion because the ear expects a tick and doesn’t get one.
Now add rim hits as syncopated accents. Try placing them at 1.1.3, 1.2.4, 1.3.2, and 1.4.4. If that feels too busy, delete one. If it feels too simple, add one, but keep it a motif.
Then add two tiny snare ghosts. Put them at 1.2.3 and 1.4.3, but keep velocity low, like 20 to 35. They should feel like little breaths, not backbeats.
Now do the unsexy part that makes it roll: velocity and micro-timing.
Hats: vary velocity roughly from 55 to 85. Add tiny random variation, two to four points, so it doesn’t feel stamped.
Rims: pick one or two and nudge them slightly late, like three to eight milliseconds. Just a touch of pocket. If you go too far, it gets drunk.
And while you listen, ask this question: does this bar sound like it’s asking something? Like it’s slightly unfinished? That’s the goal. A call that wants an answer.
Now we set up resampling, clean and repeatable.
Create an audio track called RESAMPLE PRINT.
Set its input to Resampling.
Set monitor to Off.
Arm it.
Workflow habit: solo what you’re printing. If you don’t, you’ll accidentally print half your track and wonder why your “percussion” has bass and pads baked into it. So for now, solo the DRUM SOURCE rack, or whatever percussion group you’re printing.
Loop your 1-bar call and record two to four bars. The reason we do more than one is you get little timing variations, groove differences, and you can choose the best chunk later.
Once it’s recorded, consolidate it. Select a clean region and hit consolidate, Cmd or Ctrl J. Rename it something like CALL_PRINT_174BPM_01. Be disciplined with naming. You’re about to generate multiple generations, and naming is what keeps you brave enough to commit.
Now we’re going to turn that call print into a response generator. In other words: we’re going to make new instruments out of it.
Create a new audio track called CALL to RESPONSE DESIGN, and drop your CALL_PRINT clip into it.
At this point, think in intentions. One generation per intention. Don’t build a monster chain trying to fix everything at once. Decide what you want first: do you want a tick-metal response, like gritty top texture? Or a woody shuffle, like jungle chatter?
Let’s do the tick-metal one first, because it’s a staple.
On CALL to RESPONSE DESIGN, build this device chain:
Start with EQ Eight. High-pass aggressively, somewhere between 500 and 1200 Hz. We’re carving away body so we’re left with click and edge. If you want more tick, sweep a narrow boost in the 3 to 6k area until the sound speaks.
Then Saturator. Drive four to eight dB. Soft clip on. We want density and bite.
Then Redux, sparingly. Downsample maybe two to six, and bit depth around 10 to 14. If you go too far, it turns into pure sand. Cool sometimes, but we want usable percussion, not a special effect that ruins the mix.
Then Auto Filter. Try band-pass or a high-pass with some resonance. This is your movement tool. Later you can automate cutoff so the response feels like it’s evolving.
Then Utility. Keep the width controlled, like 60 to 90 percent. The response can be wider than the call, but don’t let it smear across the whole stereo field. Remember: we want it to talk, not spray.
Set warp mode on the clip to Beats. Preserve 1/16 or 1/32. This keeps it punchy and percussive.
Now, resample this chain. This is the whole philosophy: print the sound so it becomes a real object you can edit.
Create another audio track called RESPONSE PRINT. Input Resampling, monitor off, arm it. Solo the CALL to RESPONSE DESIGN track and record one to two bars. Consolidate. Name it something like RESP_TICK_01.
Now you’ve got response material that is literally derived from the call. It will automatically feel related, even though it’s a different character.
Option two is woody shuffle, and it’s a different flavor: more body, more “instrument.”
On the same CALL to RESPONSE DESIGN track, or duplicate it if you want to keep versions, build a woody chain.
Drop a Corpus on it. Try Tube or Membrane. Tune it somewhere around 180 to 400 Hz. Keep decay short, like 0.2 to 0.6 seconds, and dry wet around 10 to 35 percent. The idea is not to turn it into a tom; it’s more like adding a resonant wood chamber behind the ticks.
Then tighten it with Drum Buss. Drive maybe five to fifteen. Keep Boom off or extremely low.
Then EQ Eight to tame any harshness. A little dip around 3 to 5k is common if it gets papery.
Then a Gate to chop the tail so it ticks and speaks rhythmically instead of washing.
Resample that as well. Record it, consolidate it, and name it RESP_WOOD_01.
Now we actually write the Response phrase. This is where a lot of people mess it up, because they think response means “more notes.” It doesn’t. Response means contrast plus logic.
Take your response print audio and build a 1-bar answer.
You can do this with pure audio editing: duplicate the clip, cut out little hits, and place them.
Or you can slice to a new MIDI track if you want the feel of playing it, but keep the mindset: it’s still resampling-based. We’re not going to external samples; we’re reorganizing our printed material.
Here’s a practical rule that works immediately.
Call bar: busy early, breath late.
Response bar: breath early, busy late.
So if your call has energy at the start of the bar, make the response light there, and then let it chatter toward the end. That creates a question-answer across two bars without you even thinking too hard about it.
Also, protect the snare pocket with a no-fly zone. In DnB, the backbeat is sacred. Keep response transients away from the exact snare hit on 2 and 4. If you want texture there, you can leave tail-only energy, but trim the transient so it doesn’t compete with the snare crack. Think about a window of about 10 to 30 milliseconds around the snare transient. Don’t collide in that moment.
Now do micro-editing like you’re editing a break.
Put tiny fades on every audio slice so there are no clicks.
Make little flams by duplicating a hit, nudging it 10 to 25 milliseconds, and lowering it 6 to 12 dB.
For mini rolls into transitions, grab a tick and place a 1/32 burst right before bar changes. That “ratchet” energy is incredibly DnB when it’s subtle.
And use clip gain like velocity editing. This is huge. Don’t rely on compression to create groove. Make 1 to 2 dB differences between hits. That’s the difference between “loop” and “performance.”
Now we glue everything together.
Group your percussion elements into a PERC BUS. Your call MIDI track, your response audio tracks, and optionally your prints if you want to keep them muted for safety.
On the PERC BUS, do basic cohesion processing.
EQ Eight first: high-pass around 150 to 250 Hz. Keep low end clean so the kick and sub are unchallenged.
Then Glue Compressor: attack three to ten milliseconds, release auto or around 0.1 to 0.3 seconds, ratio two to one. You’re aiming for one to three dB of gain reduction. Just glue, not smash.
Optionally a Saturator after for one to four dB of drive if it needs density.
And keep stereo controlled. A good rule: keep the call more centered and let the response have controlled width. Or flip it. But don’t make both wide. That’s how you get “wash” instead of conversation.
If you want that classic DnB dirt without ruining clarity, set up a parallel return called PERC CRUSH.
On that return, put Drum Buss with drive 10 to 25, then Saturator with drive six to ten, then EQ Eight high-pass 400 to 800 so only the crunchy mid-high texture returns.
Send only certain elements to it, like rims and ticks. Don’t send everything. Parallel is most powerful when it’s selective.
Now arrangement. We’re going to make it talk over eight to sixteen bars.
Try this 8-bar template first.
Bars one to four: call leads, response is minimal. Let the identity settle in. Maybe the response only does one or two little accents at the end of bar four, like it’s hinting.
Bars five to eight: full call-and-response.
Bar five: call.
Bar six: response, a bit busier.
Bar seven: call with a variation. Maybe remove two hats or shift one rim.
Bar eight: response plus a tiny fill, like a 1/32 burst into bar nine.
Then, if you extend to 16 bars, escalate with automation that isn’t just volume.
Automate brightness. Open the Auto Filter cutoff slightly on the response as you approach phrase endings.
Automate density. Mute 20 to 40 percent of hits in quieter moments, then bring them back. The track feels like it’s rising without you making it louder.
Here’s a very real DnB trick: once the conversation works, print the entire PERC BUS as audio. An 8- or 16-bar print. Name it PERC_BUS_PRINT_8bar_v1.
Then do just a few surgical break-style edits at phrase boundaries.
One reverse into a section change.
One 1/32 stutter at the end of bar eight or sixteen.
One tiny dropout, like muting one eighth note, to create a gasp of silence.
Minimal edits, maximum narrative.
A few common mistakes to avoid while you’re doing all of this.
If the response is just “more notes,” it will feel like clutter, not a reply. Contrast is the point: different density, different tone, different stereo position, different subdivision.
If you over-resample without structure, you will get lost. Use versioning. CALL_SRC_MIDI_v1, CALL_PRINT_v1, RESP_DESIGN_chainA_v1, RESP_PRINT_chainA_take1. It sounds nerdy, but it makes you faster.
Watch the low-mids. Resampled percussion loves to accumulate mud around 200 to 600 Hz. High-pass and carve. Every generation, check that area.
And don’t step on the snare. If your response hits on 2 and 4 with a similar transient, you’re weakening the backbeat. Either move it, or trim the transient so it becomes tail-only.
Now, if you want darker, heavier energy, here are a few upgrades.
Try EQ Eight in M/S mode on the response. Cut harsh highs on the sides while keeping the mid crisp. This keeps it wide but not piercing.
Make “rust” with gated reverb. Put a short, dark reverb on the response design chain, then a gate after it so it becomes a tight textured burst. Resample that tail as its own layer.
For menace, duplicate a response print and transpose it down 7 or 12 semitones. Then high-pass it at 300 to 600. Blend it very quietly. It adds weight without stealing sub space.
And if you want industrial edges, use Frequency Shifter with tiny shifts, like plus 20 to plus 80 Hz, feedback low, then resample. It creates a techy smear that still feels derived from your original call.
Let’s lock it in with a quick practice mission you can do in about 20 minutes.
Write a 1-bar call using only hat, rim, and one ghost.
Print it via resampling.
Make two responses from the same print:
one tick-metal with Redux plus Saturator,
one woody with Corpus plus Gate.
Arrange eight bars:
bars one to four, call dominates,
bars five to eight, alternate call and response each bar.
Then print the PERC BUS and do exactly one break-style edit in bar eight, like a tiny stutter or reverse into bar nine.
Your deliverable is an 8-bar percussion-only loop that could sit under a rolling sub-bass and feel like a real drum section.
Recap before you go.
You built a recognizable call from a minimal kit.
You printed it.
You mutated it into new characters using stock devices.
You wrote a response that contrasts and respects the snare pocket.
You arranged it into phrases that breathe.
And you committed to audio, which is the fastest path to that finished, cohesive DnB roll.
If you tell me your target sub-style, like roller, jungle, neuro, minimal, or dancefloor, I can suggest a specific call rhythm and a three-character response set, like tick, wood, and rust, mapped to where they hit best in that style.