Neuro Bass

The resampling-chain philosophy

Before any recipe: the loop. Make a source, process it, record it, load the recording as the new source, process it with different tools, record again. Three to five generations is normal. Keep every generation.

1. Set up a template: a MIDI track with the source synth, an audio track with input Resampling, and a Return with a dub delay. Name audio clips "neuro_gen1_notch14", "gen2_shifter" etc. the moment they are recorded.
2. Each generation uses a different modulation rate and a different type of processing (distortion → spectral → pitch → rhythmic). Repeating the same tool twice just makes it louder.
3. Always record 8 bars with the modulation running — long enough that slow rates (2 bars, 4 bars) complete a cycle.
4. High-pass at 150 Hz before every distortion stage. The sub is re-added from a sine at the very end, driven by the same MIDI as the final chop.
5. In 12.2+ Bounce in Place on a MIDI clip does generations 1 and 2 faster than manual resampling; Bounce Group (12.3) renders a whole group of layers in one go.

FM Operator source → Roar → Auto Filter → Shifter → resample

The canonical generation-1 chain. An FM source with inharmonic sidebands, a serial saturator to multiply them, a tempo-synced notch to organise them, and a frequency shifter to smear them.

1. Operator, serial algorithm. A saw carrier. B sine ratio 3.5 (Coarse 3, Fine +50) at −12 dB. C sine Coarse 7 at −24 dB feeding B. Operator LFO synced 1/4, sine, to B level ±40% — the sideband density pulses every beat.
2. EQ Eight HP 150 Hz 24 dB before Roar. Non-negotiable.
3. Roar Serial: Soft Sine 8 dB → Fold 5 dB. Post filter LP 7 kHz. The fold stage mirrors sidebands above the fold threshold back into the mids — more inharmonic content.
4. Auto Filter: Notch, resonance 50%, frequency 900 Hz, LFO synced 3/8 (dotted quarter), amount 60%, triangle. 3/8 against the Operator's 1/4 resolves every 3 beats — already two rates.
5. Shifter in Frequency (Hz) mode, not Pitch: +30 Hz shift, Dry/Wet 40%. Its LFO at 1/16 ±60 Hz. A frequency shift (as opposed to pitch shift) makes harmonics inharmonic by a constant Hz offset — the signature neuro "ring".
6. Record 8 bars to the Resampling track. This is generation 1.

Spectral Time freeze and Spectral Resonator on generation 1

Generation 2 moves into the spectral domain. Spectral Time's freeze creates sustained spectral snapshots that are retriggered rhythmically; Spectral Resonator pitches the mess back into key.

1. Simpler with generation 1, Classic mode, Warp on so pitch follows MIDI while the movement stays in tempo. Play the bassline (root notes with a fifth jump).
2. Spectral Time: Freeze section on, Retrigger synced to 1/8 — every eighth note a new spectral snapshot is captured and held. Fade In 20 ms so the freezes do not click. The Delay section at 3/16 with 40% feedback adds spectral echoes between retriggers.
3. Tilt the frozen spectrum slightly upward (+2) to keep brightness through the freeze.
4. Spectral Resonator in MIDI mode receiving the same MIDI track: Decay 250 ms, Harmonics around 50%, Stretch +8% (slightly inharmonic partial spacing — more metallic). 35% wet. This re-tunes the frozen chaos to the note.
5. EQ Eight: HP 150, and a gentle −3 dB at 2.8 kHz where spectral processing tends to pile up resonances.
6. Record 8 bars. Generation 2.

Pitch Hack and Grain Delay texture pass

Pitch Hack is a pitch-shifting delay with reverse and recycle; Grain Delay is a granular pitch-shifting delay. Both add pitched copies of the bass displaced in time — the "stuttering harmonics" of modern neuro.

1. Pitch Hack: Delay time synced 1/16, Pitch +12 semitones, Reverse 30% (that proportion of delay repeats play backwards), Recycle 25% (repeats are re-pitched again), Dry/Wet 30%. Automate Pitch between +12 and −12 on the last bar of a phrase.
2. Grain Delay: delay 1/8T, Spray 15 ms (random grain timing), Pitch +7 semitones, Random Pitch 10%, Feedback 20%, Dry/Wet 25%. The fifth above adds a "sung" ghost; the spray makes it granular rather than a clean echo.
3. Roar Single Diode 6 dB afterwards: distorts the pitched copies together with the dry signal so they fuse into one texture rather than sounding like effects.
4. Record generation 3. The bass now contains fifths and octaves of itself, reversed and granulated, locked to 1/16 and 1/8T.

Corpus comb and phase/notch movement

Comb filtering is the hidden engine of neuro: a short delay mixed with the dry signal creates a series of evenly spaced notches, and moving the delay time sweeps all of them at once. Corpus in Tube/Pipe mode, Phaser-Flanger in Flanger mode and a Doubler-style Delay all do this with different flavours.

1. Corpus, Pipe resonance type, MIDI frequency from the track so the comb spacing follows the note. Decay 20% (short — we want the comb, not the ring), Material 50%, Dry/Wet 50%. Corpus LFO on Tune, 1/4 sync, ±2 semitones: the comb sweeps a tone every beat.
2. Phaser-Flanger, Flanger mode: delay 2 ms, Feedback 60%, modulation LFO at 1 bar, wide depth. Positive feedback emphasises the comb peaks, negative feedback the notches — try both; negative is the more "neuro" sound.
3. EQ Eight: three narrow notches (Q 6) at 700 Hz, 1.6 kHz and 3.4 kHz, −12 dB. Static notches in a moving spectrum sound like a filter moving the other way — a cheap extra "movement" layer with zero modulation.
4. Map an LFO (M4L) to the EQ Eight's Scale parameter (which scales all gains) at a 2-bar rate to make the static notches breathe.

"Neuro reese to growl" morph

The classic neuro phrase: a smooth reese that over 2–4 bars tears open into a growl. Done in one Wavetable with the mod matrix driving unison, position, FM and filter from a single macro or a 4-bar envelope.

1. Osc 1: Basics saw, Unison Phase Sync 4 voices, Amount 15%. Osc 2: Formant table, level 0 (off) at the reese end. Sub off.
2. Osc 1 effect mode FM, amount 0 at the reese end. Filter 1 Notch MS2 res 40% at 600 Hz; Filter 2 LP OSR 1.2 kHz; routing Serial.
3. Put Wavetable in an Instrument Rack and make one macro "Morph". Map, with Live 12 macro ranges: Osc 2 level 0 → −6 dB; Osc 1 FM amount 0 → 45%; Unison Amount 15 → 40%; Filter 2 cutoff 1.2 → 5 kHz; Filter 1 freq 600 → 1800 Hz. Set the mapping curves (right-click the macro mapping) so Osc 2 arrives late and FM arrives early.
4. Mod matrix inside Wavetable: LFO 1 (1/8) → Osc 2 Position 40%, LFO 2 (3/16) → Filter 1 Freq 30%. These only become audible as the morph brings Osc 2 and the notch range in.
5. Roar Serial after: Hard 6 dB → Fold 4 dB. Map Roar Dry/Wet to the same Morph macro, 0 → 100%.
6. Automate Morph: flat at 0 for bar 1, ramp to 70% over bars 2–3, snap to 100% on bar 4 beat 3. Resample it.

Wavetable mod-matrix bass with stacked rates

A single-device neuro source using all four Wavetable modulators at different divisions. Not as deep as resampling, but it gives a generation-0 source that already has four rates of movement.

1. Osc 1: a Distortion-category table (these are pre-folded). Osc 2: Harmonics-category table at −1 octave, 50%. Sub on at −12 dB (temporary — mute it before resampling).
2. Filter 1: MS2 BP res 35%. Filter 2: PRD (the Prophet-style model) LP with Drive 20%. Routing: Parallel so the two filters give two resonant peaks — a two-formant mouth.
3. Mod matrix: LFO 1 sync 1/4 → Osc 1 Position 50%. LFO 2 sync 3/16 → Filter 1 Freq 40%. Env 2 (A 0, D 300 ms) → Osc 2 Position −60%. Velocity → Filter 2 Drive 40% so accents distort more.
4. Set LFO 1 shape to the Wavetable "steps" or sample-and-hold style option and LFO 2 to triangle — one stepped movement, one smooth one.
5. Roar Multiband: low band clean, mid band Fold 6 dB, high band Soft Sine 4 dB. Then EQ HP 150 and Bass Mono.

Beat Repeat glitches and rhythmic edits

Beat Repeat on a resampled bass generates the stutters, reverse-feeling pitched repeats and half-beat drop-outs that are impossible to program by hand at speed. Record its output; keep the best bars.

1. Beat Repeat: Interval 1 bar (how often it may trigger), Offset 3/16 (where in the bar), Chance 60%. Grid 1/32 for machine-gun stutters, with Variation set to 4 and Variation mode Random so the grid wanders between 1/32 and 1/8.
2. Gate 1/8: each repeat burst lasts an eighth. Pitch −12 with Pitch Decay 40%: repeats slide down an octave over the burst — the classic neuro "drop" edit.
3. Mix mode: Insert (dry is muted during repeats) for clean edits; Gate mode (only repeats are heard) for building sparse fill tracks.
4. Record 16 bars, audition, crop the best 4 bars into a clip. Use the clip's Live 12 Audio-to-MIDI (convert to a new MIDI track) only if you want the rhythm, not the audio.

Tempo-synced modulation hierarchy

Rather than random rates, assign each generation a rate from a planned hierarchy. Slow rates on tonal parameters, fast rates on textural ones, and at least one non-power-of-two rate per generation.

Mastering the mid-range (1–4 kHz) and mono checks

Neuro basses are won or lost between 1 and 4 kHz: too much and it is fatiguing, too little and the growl disappears under the drums. This is the mix pass before the final sub re-add.

1. Multiband Dynamics with the mid band set to 1–4 kHz: downward compression 3:1 above −14 dB, Attack 1 ms, Release 40 ms. This catches the resonant peaks from notches and combs without dulling the whole bass.
2. EQ Eight: a broad +2 dB at 2 kHz (Q 0.8) for presence, −2 dB high shelf above 7 kHz to stop the Shifter/Grain fizz competing with hats.
3. Check against the snare: solo bass + snare. The snare's body (180–250 Hz) and crack (2–5 kHz) must still read. If not, dynamic-cut the bass at those bands with the snare as sidechain input on Multiband Dynamics.
4. Mono check: Utility on the master, Width 0%, listen for level drop and for any hollowness in the mids — if the mids change in mono, something upstream has stereo phase tricks that need to go. The Live 12.3 A/B compare on the Utility makes this a one-click toggle.
5. Add the sub sine now: separate track, same MIDI as the final chop (convert the chop clip's slices to MIDI notes, or write it). LP 100 Hz, mono, sidechained to the kick.

An 8-step recipe for a complete neuro patch

The condensed procedure, start to finish. Budget an hour.

1. Source. Operator FM (saw carrier, ratio 3.5 and 7 modulators) or the Wavetable mod-matrix patch. Mono, 8-bar clip of root notes with one fifth jump. HP 150 Hz.
2. Generation 1 — distortion + notch. Roar Serial (Soft Sine → Fold), Auto Filter notch at 3/8, Shifter 1/16 at 40% wet. Resample.
3. Generation 2 — spectral. Simpler (Warp on) → Spectral Time freeze retrigger 1/8 → Spectral Resonator MIDI mode. Resample.
4. Generation 3 — pitch/time texture. Pitch Hack 1/16 +12 reverse 30% → Grain Delay 1/8T +7 → Roar Diode. Resample.
5. Generation 4 — comb and static notches. Corpus Pipe (MIDI tune, LFO 1/4) → Phaser-Flanger Flanger → EQ Eight 3 notches. Resample.
6. Edits. Simpler Slice mode on generation 4 or Beat Repeat pass; write the final 4-bar bassline as slice MIDI with pitch drops and reversed slices.
7. Mid-range mix. Multiband Dynamics 1–4 kHz, EQ presence/fizz shaping, Bass Mono 150, mono check, light kick sidechain.
8. Sub re-add. Operator sine on a separate track, same MIDI, LP 100 Hz, hard sidechain. Glue both on a bass group bus.

Generating variations with Macro Variations and randomize

Once a generation-0 synth rack exists, Live 12's macro tools can produce dozens of sources for resampling in minutes, and the variation index itself is automatable.

1. Build the rack with 8 macros covering the generation-0 parameters. Set each macro's min/max to the musically useful range so random values are never useless (e.g. Fold 0–8 dB, not 0–30).
2. Click the randomize (dice) button in the rack's macro section. Exclude macros you never want randomised (right-click → exclude from randomisation): Sub level, Width.
3. When a result is good, store it as a Macro Variation. Build 8 variations: Clean, Reese, Growl A, Growl B, Hollow, Fizz, Comb, Broken.
4. Automate the variation selection per bar in a clip, or map the Variation launch to Push pads. Record 4 bars of each, or one 32-bar clip cycling through them, as generation 1 material.
5. Live 12 also lets Rack snapshots be recalled via MIDI mapping — in a live set this is how a neuro bass "changes patch" every 8 bars without a second track.