ABLETON 12 SOUND DESIGN ATLAS
Home › Fundamentals

Routing and Resampling

Live's track I/O in full — Audio From/To, Pre FX/Post FX/Post Mixer taps, sends and returns, groups, sidechain inputs, monitoring modes, Resampling, External Instrument and External Audio Effect, dummy clips, return feedback, Bounce (12.2/12.3), Freeze/Flatten and latency compensation, with diagrams of the setups that come up constantly in sound design.

routingresamplingaudio fromaudio tosendsreturnsgroupssidechainexternal instrumentexternal audio effectpost fxpost mixerdummy clipsfeedbackmonitoringbouncefreezeflattenlatencydelay compensation

Most of what gets called "advanced sound design in Live" is routing. The synth is the same; what changes is where you tap its signal, what you feed back into it, and how you print the result. Any track can take audio from any other at three points (before devices, after devices, after fader), returns can feed each other, and a Resampling input records whatever the master plays. External Instrument and External Audio Effect pull hardware into the same graph. This page maps that graph, then gives the setups that come up in every session: parallel chains, sidechain taps, dummy clips, feedback loops, hardware inserts and the three ways of printing.

Architecture / concept

Open the In/Out section with Cmd+Opt+I. Every audio track has an Audio From chooser (Ext. In, Resampling, any track or return, No Input) with a sub-chooser for the tap point, a Monitor switch (In / Auto / Off) and an Audio To chooser (Master, Sends Only, Ext. Out, another track). Groups sum their children; returns have only Audio To and send knobs. The taps:

TapWhereUse
Pre FXTrack input, before any device (for MIDI tracks: the instrument output)Parallel chains from the raw source
Post FXAfter the device chain, before fader and panSidechain keys that must not change when you mix; parallel chains
Post MixerAfter fader, pan, mute and sendsRecording what you hear; stems
Rack chains and drum padsAny rack chain, any Drum Rack padTapping one layer of a layered instrument

Sends, returns and groups

Sends are post-fader by default (Pre/Post toggle per return). Returns can send to other returns and, once enabled by right-clicking the send knob, to themselves; both are off by default because they create feedback paths. Groups hold devices, so a group is a free bus, and since 12.3 can be bounced to one audio track. Sidechain inputs on Compressor, Gate, Multiband Dynamics, Auto Filter, Roar (12.2), Auto Shift, Vocoder and the M4L Envelope Follower take any track at any tap, a path that never passes through the mixer.

Monitoring modes

ModeBehaviourUse
InAlways passes the input through the devices, ignores clipsHardware returns, dummy clip hosts, live input processing, feedback loops
AutoInput while armed, clips when notNormal recording
OffNever passes input; records it silently when armedResampling and tracking where you monitor the source elsewhere

Resampling, Bounce, Freeze and Flatten

Resampling is an input choice: Audio From → Resampling, arm, record, and the track captures the master output in real time, including everything you do by hand. Freeze renders a track with its devices to a hidden file and disables them; Flatten replaces the track with that audio. Bounce in Place (12.2) renders selected clips or a track with its devices to audio, offline and faster than real time; Bounce Group (12.3) does it for a group and its children in one go. Resample when the performance matters; Bounce for a clean, phase-aligned print; Freeze when you only need CPU back.

Hardware I/O devices

External Instrument sits on a MIDI track: MIDI To (port, channel), Audio From (interface inputs), Gain and a Hardware Latency field. External Audio Effect is the audio equivalent with Audio To and From, Dry/Wet, Gain, the latency field and a measurement button that pings the round trip. Both make a hardware box behave like a plug-in inside the compensated graph, and both can be frozen and bounced, which is how you print a eurorack voice or a TD-3 pass. See hardware and modular integration.

Latency compensation

Delay Compensation (Options menu) delays every path so the one with the most device latency arrives with the rest; hover a device title bar to see its latency (lookahead limiters, linear-phase EQ and convolution add several ms). Routing a track into another is compensated; feedback loops cannot be, so a return feeding itself always carries at least one buffer of delay (128 samples at 48 kHz is 2.7 ms). Reduced Latency When Monitoring removes compensation from Monitor In tracks so live inputs feel immediate, at the cost of drifting from the compensated ones. The Track Delay field nudges a track by hand for phase-aligning layers or offsetting hardware.

Parameters that matter

WhereParameterWhat it doesStarting points
Track I/OAudio From + tapSource and tap pointPost FX for sidechains and parallel chains; Post Mixer for stems
Track I/OAudio ToMaster, Sends Only, Ext. Out, or another trackSends Only for a track that exists only to feed returns
Track I/OMonitorIn / Auto / OffIn for anything that has no clips and must pass audio
ReturnPre / Post toggle, send-to-self enablePre-fader sends; self-feedback permissionPre for dub sends you fade out while the tail stays; self-send off unless building a feedback rack
External Audio EffectAudio To/From, Gain, Dry/Wet, Hardware Latency, MeasureHardware insert with compensationMeasure once per interface and sample rate; Gain to match the hardware's input level
External InstrumentMIDI To, Audio From, Gain, Hardware LatencyHardware synth as instrumentLatency 2–8 ms typical over USB; Gain +0 to +12 dB for −10 dBV gear
Options menuDelay Compensation, Reduced Latency When MonitoringGlobal compensation and the live-input exceptionBoth on for tracking; Reduced Latency off when mixing
Audio preferencesBuffer size, Driver Error CompensationRound-trip latency and its measured correction128 samples for playing, 512+ for mixing; run the compensation test with a loopback cable
FreezeFreeze / Flatten, UnfreezeRender in place, replace with audio, restoreFlatten only on a duplicate of the track
Bounce (12.2+)Bounce in Place, Bounce to New Track, Bounce Group (12.3)Offline render of clip, track or group with devicesBounce to New Track to keep the source muted below for later edits

Techniques

Parallel chain from a Post FX tap

Devices: track routing, Roar, EQ Eight · Level: beginner · Genres: techno, dnb, industrial

The simplest parallel processing in Live is a second audio track listening to the first. Unlike an Audio Effect Rack with two chains, the two paths have their own faders, sends and automation lanes.

  1. New audio track, Audio From the bass track, sub-chooser Post FX, Monitor In.
  2. Distortion on the parallel track, then an EQ that removes the sub (HP 150–200 Hz) so the clean sub comes only from the main track.
  3. Balance with the fader; automate the parallel fader for drops where the grit comes forward.
Both paths are delay-compensated as long as Delay Compensation is on, so there is no phase smear between the clean low end and the distorted mids. See parallel processing.
Pre FX bypasses the main track's devices for the parallel path: sometimes wanted (raw source for a crushed layer), sometimes a surprise.

Sidechain taps that survive the mix

Devices: Compressor, Gate, Auto Filter, Roar sidechain · Level: beginner · Genres: house, techno, dubstep

A sidechain keyed from Post Mixer changes whenever the kick fader moves. Key from Post FX or from a silent trigger track and the ducking is independent of the mix.

  1. Duplicate the kick track, name it Ghost Kick, set Audio To → Sends Only, all sends down. It now makes no sound.
  2. Point every sidechain (Bass Compressor, Pad Auto Filter, Roar on the chords) at Ghost Kick, Post FX.
  3. Now the ghost clip can play 4-to-the-floor through a breakdown where the real kick is out, keeping the pump; or play a different rhythm from the audible kick.
Live 12's Gate, Auto Filter (12.2 redesign) and Roar (12.2) all have sidechain inputs with Pre/Post taps. The Envelope Follower from Max for Live modulators extends the same idea to any parameter.

Dummy clips: clip envelopes as a performance effect rack

Devices: audio track with Monitor In, Beat Repeat, Auto Filter, Echo · Level: intermediate · Genres: performance, techno, dub

An audio track that listens to another track and holds silent clips whose envelopes drive its devices. Launch a clip and the effect pattern changes while the source keeps playing through it.

  1. Audio track FX Host: Audio From Drums, Post Mixer, Monitor In. Set the Drums group Audio To → FX Host (Track In) if you want the drums to be heard only through the host; otherwise keep both and treat the host as parallel.
  2. Drop any short audio clip into the host, set its clip gain to −inf (or use an empty silent file). Loop 1 bar.
  3. In the clip's Envelopes tab, draw Auto Filter Freq, Beat Repeat Chance, Echo Dry/Wet. These lanes can be automation or modulation; modulation leaves the knobs free for your hands.
  4. Duplicate the clip for each variation: "filter sweep", "stutter", "dub out". Launch them like scenes.
Clips carry envelopes and Monitor In tracks pass audio regardless of clips; together that is a pattern-launchable effects processor with no M4L needed. See performance.

Send-to-self feedback dub delay

Devices: return track, Echo or Delay, EQ Eight, Limiter · Level: intermediate · Genres: dub, dub techno, no-input

Enable a return's send to itself and the delay output re-enters through whatever devices sit in the loop: filter, saturate, pitch-shift, so the repeats evolve instead of just decaying.

  1. Right-click the send A knob on return A → Enable Send. Set it to −inf for now.
  2. Delay with device feedback 0 and Dry/Wet 100%. EQ Eight with HP 200 Hz and LP 5 kHz in the loop so each repeat gets darker and thinner. A Limiter at the end of the chain, not in the loop's feed, as the fence.
  3. Raise send A on return A slowly. Around −12 to −6 dB the repeats hold; above −3 dB the loop runs away (the limiter stops the damage but the tail is now a drone, which can be the point).
  4. Map the self-send knob to a macro or a Push encoder; this is the dub "send to infinity" move.
The loop runs through one audio buffer of delay that Delay Compensation cannot remove, so the effective delay time is the device's time plus the buffer. At 1024 samples that is 21 ms, audible as a timing drift on synced repeats; use 128–256 samples when performing feedback sends.

Variations: Shifter at +12 st in the loop for a rising shimmer that climbs an octave every repeat; Grain Delay with Spray in the loop for a cloud; Roar's own Feedback routing does a device-level version for shorter, pitched loops. See dub FX and delays and the feedback rack.

Resampling a performance into a Simpler layer

Devices: Resampling input, Simpler, Drum Rack · Level: beginner · Genres: all, especially techno, idm, lofi

Record the master while riding macros, then chop the result as a sample. The devices are gone; the gesture is baked in.

  1. New audio track, Audio From → Resampling, Monitor Off, arm. Solo the source track. Bypass the master limiter so you print headroom, not a squashed take.
  2. Record while performing the macros: filter rides, the feedback send above, a rate switch on a wobble LFO.
  3. Drag the clip onto a Simpler in Slice mode (Transients) or into a Drum Rack. Each slice is now a hit with the movement inside it.
  4. Repeat: resample the Simpler track again through a different chain. Two or three generations of this is where most interesting textures come from.
Launch recording on the bar and the slices land on the grid without warping. Deeper: resampling workflow and the resampling rack.

Hardware insert on a Live track (External Audio Effect)

Devices: External Audio Effect, eurorack filter, pedals · Level: intermediate · Genres: techno, electro, industrial

Treat a eurorack filter, a pedal or a hardware compressor like a plug-in with compensation. The device replaces the pair of "send out, record back in" tracks and keeps everything on the grid.

  1. External Audio Effect on the track. Audio To → the interface outputs that go to the hardware, Audio From → the inputs it comes back on.
  2. Press the latency measurement button with the hardware in bypass or with a cable loop, then put the hardware back in circuit. Re-measure if you change buffer size or sample rate.
  3. Gain to match levels (eurorack is hot, pedals are not; −20 to +12 dB range). Dry/Wet below 100% gives a free parallel blend.
  4. Freeze or Bounce the track when you are done so the hardware is free for the next job and the set is portable.
Without the device, a Monitor In return is either excluded from compensation or included un-measured. The device gives a measured number and keeps the return inside the graph.

Printing: Bounce, Freeze/Flatten or Resample, and when

Devices: Bounce in Place (12.2), Bounce Group (12.3), Freeze · Level: beginner · Genres: all

Three ways to turn a processed track into audio, each with a different relationship to time, latency and tails.

MethodSpeedCapturesMissesUse
ResampleReal timeMaster output, your hands, any live input, feedback loopsRequires solo discipline; prints the master chainPerformances, hardware, feedback, "happy accidents"
Freeze / FlattenOfflineTrack devices, clip automation, tails (Freeze adds tail time)Sidechain inputs from frozen tracks are static; nothing from other tracksCPU relief, keeping the source editable (Unfreeze)
Bounce in Place (12.2)OfflineSelected clips or track with devices and automation, to a new or same trackLive inputs and feedback paths; Monitor In tracksCommitting a sound design pass, phase-clean stems
Bounce Group (12.3)OfflineWhole group with group devices to one audio trackSame as aboveDrum groups, layered synth groups, delivering stems
Bounce and Freeze both respect Delay Compensation, so the printed audio lines up with the originals to the sample. Resampling also does, as long as you monitor through the master rather than a Reduced Latency input.
A frozen track still feeds sidechains and Audio From taps with its frozen audio, but a track that receives from a frozen track via Audio From is not frozen with it. Freeze from the bottom of the chain upward when a routing graph is involved.

Drum Rack pads to individual tracks and back

Devices: Drum Rack, return chains, group · Level: beginner · Genres: all drum programming

Drum Rack chains can be routed to their own tracks from inside the rack, extracted (right-click a pad → Extract Chains), or tapped via another track's Audio From. The kick gets its own compressor and the hats their own send without leaving the rack.

  1. Show the Drum Rack's chain list and press its I/O button. Each chain gets an Audio To chooser; route Kick, Snare and Hats to three audio tracks set to Monitor In.
  2. Group the three out tracks plus the Drum Rack track into a Drums group with Drum Buss and Glue on the group.
  3. Sidechain the bass from the Kick Out track Post FX, not from the rack, so the kick's own compressor and EQ are part of the key.
The rack keeps MIDI, pad layout and macros in one place for Push; the out tracks give each element a mixer channel. See Simpler, Sampler and Drum Rack.

Related