Reverb and compression (offline, in the sample editor)
What it is#
A tracker mixes samples dry. Room, depth and punch therefore have to be baked into the samples beforehand in a sample editor (Airon's examples are Cool Edit and Sound Forge), or faked inside the tracker with extra channels (fake-echo) and envelopes (instrument-envelopes). Airon's 1998 tutorial gives the parameters, numbers and two recipes (source: s-tw118-reverb-and-compression); Rimbo adds monitoring advice (source: s-tw119-reverb-and-audio-gear).
Why it works#
- The brain reads room size from the early reflections — the first few echoes off the nearest walls — and room "warmth" from how fast highs die: a crowded, wooden room absorbs them quickly, a hard empty one keeps them and sounds "emptier and colder", small hard rooms "very metallic" (source: s-tw118-reverb-and-compression).
- The attack identifies a sound. A compressor whose attack time is shorter than the sample's own attack crushes the transient — which is why the attack setting is the one number to get right, and why the same mistake on a snare is usable as an effect (same source).
- A reverb also acts as a frequency lens: "the right reverb at the right mix can really coax the best frequencies out of a sample" where EQ fails (source: s-tw119-reverb-and-audio-gear).
How to do it#
Sample editor (Cool Edit / Sound Forge): reverb#
Parameters, in the order the sound arrives (source: s-tw118-reverb-and-compression):
| Parameter | Meaning | Club example (8 × 8 × 3 m, band 4 m away) |
|---|---|---|
| Dry / wet | the direct sound / everything after it | wet modest — loud reverbs need big hard rooms (a church), not a club |
| Early reflections | first echoes off the nearest walls; the room-size cue | ≈15 ms: 9 m path ÷ 343 m/s = 26 ms, minus the 11 ms direct path the dry sample already covers (Sound Forge offers a list; Cool Edit has no ER control) |
| Pre-delay | time from direct sound to the reverb peak | 30 ms or more; larger rooms develop reverb later |
| Reverb time | decay to −60 dB | under 1 s (try 0.7–1.2 s) for a small crowded room |
| Density | how dispersed the tail is (marble reflects, wood disperses, carpet swallows) | high — people, wood, obstacles |
| High-frequency absorption | how fast highs die (Cool Edit High Frequency Absorption Time; Sound Forge Attenuate high frequencies above) | absorption time very low / attenuate above 2000–2500 Hz |
Placement by amount of reverb: snare to the front, piano a little back, sax just in front of the drums; "Snare always gets reverb". Decide first whether you simulate a live band or a multitrack production (doubled takes, extreme panning — "NOT reality") (same source).
Sample editor: compression#
The signal passes through a compressor ("DRY goes in, WET comes out"): above the threshold it is reduced by the ratio, after the attack time, and it stops being reduced a decay (release) time after the signal drops back; an infinite ratio is limiting. Settings per sound (source: s-tw118-reverb-and-compression):
| Sound | Threshold | Ratio | Attack | Decay | Notes |
|---|---|---|---|---|---|
| Synth sample that will not come forward | −6 dB | 4:1 | just longer than the sample's attack (14 ms attack → 15 ms) | ≥500 ms for a single hit; 20–30 ms for a sequence of sounds | normalise afterwards |
| Bass drum | — | — | ≥30 ms | ≥80 ms | compress first, then boost the bass; never touch the attack |
| Acoustic guitar | −3 dB | 2:1 (–3:1) | — | — | "as little as possible" |
| Snare as an effect | — | — | short enough to crunch part of the attack | — | changes the snare "considerably" |
| Vocals (studio reference) | not too low | 2:1–6:1 (radio hosts 8:1) | — | — | peaks over 100 dB SPL vs ~86 dB tape dynamic range |
Recipe: gated reverb snare#
- Normalise the snare, copy it to a new window; everything below is done on the copy.
- Reverb with dry = 0 and the high-frequency cutoff/absorption set high so the tail keeps its highs; normalise.
- Compress: threshold −12 dB or lower (−13, −14 …), attack 1 ms, decay 500 ms, ratio 12:1 or more; normalise again.
- Fade up the first ~30 ms (Sound Forge's fade function, Cool Edit's envelope).
- Mix: original at 70–80 % (at least −3 dB — its attack must dominate) plus as much of the wet copy as you like.
The 1 ms attack and the fade-up replace the noise gate that gives the effect its name (source: s-tw118-reverb-and-compression).
Recipe: preverb (reverse reverb)#
- Normalise and copy the sample.
- Insert silence of about 90 % of the sample's length at the start of the original.
- Reverb the copy with dry ≈ 0 and a reverb time of at least 1 s (or a big-hall preset — turn its dry level back down).
- Reverse the reverbed copy and mix it into the silent start of the original.
"Nothing but a reverse reverb put in front of the original sound" — cannot be done with outboard hardware, "great ghost stuff" (source: s-tw118-reverb-and-compression).
Inside the tracker#
The channel echo of the 1990s scene is, in Airon's words, "a primitive version" of reverb — its early reflections (source: s-tw118-reverb-and-compression); see fake-echo for the copies-on-extra-channels method, instrument-envelopes for looping a reverb tail under a volume envelope, and stereo-panning for spreading the echoes of a reverb-treated sample.
Tips & pitfalls#
- "NEVER DRIVE THE SAMPLE INTO CLIPPING" — digital clipping is not analog overdrive and will not give you a distorted guitar (source: s-tw118-reverb-and-compression).
- Normalise after every compression pass; compress a kick before boosting its bass (same source).
- Before reaching for more EQ, try a room-sized reverb at a low mix — it may bring out the frequencies you wanted (source: s-tw119-reverb-and-audio-gear).
- Monitor on both: headphones are the cheapest accurate reproduction, but re-mix on speakers — fuller bass, duller treble, less extreme panning, "things just sound different". Buy monitors before headphones (about $400–500 in 1998), from a pro-audio store: home audio is built to sound good, pro audio to be accurate (source: s-tw119-reverb-and-audio-gear).
- Mix checks (Zinc, 1997, from Future Music's studio tips): listen in mono, on bad speakers, at very low and very high level, from the corridor and from different spots in the room; no Loudness or bass boost on the monitoring hi-fi; bass in the centre, drums near it, strings and guitars wider; don't normalise to 100 % if the song is meant to be quiet; save samples dry and add effects later; and "Garbage In, Garbage Out. There's no such thing as 'hiding it in the mix'" (source: s-tw096-studio-tracking-tips; the habits side of the same list is on tracking-workflow). Atlantic's older rule for the same problem: keep a frequency spread — low and high sounds at once give depth (source: s-tw032-not-so-advanced-tracking-tips).
- On chips there is no reverb at all; the same jobs are done with fake-echo, retrigger and volume tables — see chiptune.
Related#
fake-echo · instrument-envelopes · stereo-panning · instrument-design · recording-samples · chip-samples · realism-in-tracked-music · tracking-workflow · levels-and-headroom
Sources#
s-tw118-reverb-and-compression · s-tw119-reverb-and-audio-gear · s-tw096-studio-tracking-tips · s-tw032-not-so-advanced-tracking-tips