Roar: choose where colour happens, then match the level

One stage, one listening question

The first example uses an original eight-second synthesized bass with uneven accents, not a recorded instrument. Routing is Single, Shaper 1 is Soft Sine, Drive is about +6 dB, Amount is 20%, Tone is 0%, Comp is 0%, Output is about −12 dB and Dry/Wet is 100%. Start at a quiet playback level. Follow the quieter notes as well as the strongest accent: does the added edge make their pitches easier to hear?

Drive raises the signal entering the stages; Amount changes the selected shaper's strength. They are not interchangeable volume controls. Lower Drive toward 0 dB, then restore it and lower Amount toward 10%. Change one at a time and listen for the balance of note body, upper harmonics and accent contrast. If every note becomes equally insistent, reduce the processing rather than adding more bass.

Compare Dry/Wet at 0% and 100%, adjusting Output while fully wet until the two versions feel similarly loud. The example's output cut is only a starting point, not measured loudness compensation. Output follows the stages and compressor and affects the wet signal, not the original dry signal. Lowering it does not undo saturation; lowering Drive changes the sound that produced that level.

Choose a curve before correcting its tone

With the bass fully wet, compare Soft Sine with Diode at Amount 20%. Recheck Output after switching. Ask whether the new roughness helps the note or distracts from it. The menu also includes Digital Clip, Overdrive, Saturated, Tube, Bit Crusher, Gnarly, Noise Inject, Shards and Hard Curve. This is an analytic approximation: several names currently share a curve family, so do not expect eleven distinct behaviours or exact Live 12 timbres.

Return to Soft Sine and try Tone at −30%, 0% and +30%. In this implementation positive Tone boosts the low side and reduces the high side; negative Tone does the reverse. The tilt follows the saturation stages, with a default corner of 180 Hz. Use it to rebalance the result, not to decide which frequencies reach the shaper. If a darker sound is still too flattened, reduce Drive or Amount instead.

Compression is a separate decision

Raise Comp from 0% to 30% while listening to the strong note and its decay. Comp controls an output RMS leveller, not another shaper. Its threshold and timing are fixed rather than exposed as panel controls. A detector high-pass is on by default, reducing low-frequency influence on gain reduction without cutting the audible bass itself. Match Output again; return Comp toward zero if the accents lose too much contrast.

Let routing decide what gets coloured

The second original eight-second source alternates low kick-like tones with noisy snare-like hits. This preset selects Multiband, with crossover settings of 200 Hz and 2000 Hz. Only stage 2 is enabled, using Diode at 25%; stages 1 and 3 pass their bands unchanged. Drive is 0 dB, Tone 0%, Comp 30%, Output about −6 dB and Dry/Wet 40%. The goal is middle-band grit beside the original rhythm, not a louder drum loop.

The compact panel exposes Routing, Shaper 1, Drive, Amount, Tone, Comp, Output and Dry/Wet. Shaper 1 and Amount address stage 1 only, so they do not change this preset's active middle stage. The example supplies the additional stage settings; it does not reveal a second-stage panel. Compare Dry/Wet at 0%, 40% and 100%, then try a small Drive change. Listen for the noisy hit's texture and whether the low pulse remains useful.

Single uses stage 1; Serial feeds enabled stages in order. Multiband splits low, middle and high content across stages 1, 2 and 3 before summing. Its gentle crossovers are not brick-wall isolation. Parallel and Mid/Side appear in Routing but currently use the Serial path: they do not provide independent parallel branches or centre-versus-side saturation. In this drum preset, switching to Serial sends the full signal through stage 2, making it a useful contrast with Multiband.

Know what the wider engine does

The stored effect also supports each stage's enable switches and post-shaper filter type, frequency and resonance, plus tone frequency, crossovers and feedback amount, time, frequency and polarity. These are not all knobs in the compact panel. Feedback routing returns a filtered, delayed signal into the stages when its amount is above zero; choosing Feedback alone with the default zero amount adds no loop. Keep feedback low when exploring it, since recirculation can build level quickly.

Do not plan around Live's full noise and matrix-modulation system here. Noise Inject is currently a shaper label, not a separate working noise generator; there is no modulation matrix in this implementation. Stage Bias is stored but is not applied by the current kernel. Finish with a moderate wet blend and a level-matched comparison in context. Roar's compressor is not a final peak limiter, and adding stages is useful only if the musical detail survives.