Parallel compression blends a heavily compressed duplicate of a signal with the original dry signal running simultaneously. You get the density, sustain, and glue of aggressive compression without sacrificing the snap and attack of natural transients — the best qualities of both processed and unprocessed sound in a single blend.
What Is Parallel Compression?
Parallel compression — sometimes called New York compression or upward compression — is a signal processing technique where the original, uncompressed signal and a separately compressed version of the same signal are combined at the mix bus or channel level. The resulting blend is richer than either signal alone.
Standard serial compression processes the whole signal. Every peak gets reduced; every quiet moment gets proportionally louder. The problem is that the compressor's attack time, no matter how fast, always imposes some degree of gain reduction at the moment of the transient. Set it too aggressively and you lose the crack of the snare, the punch of the kick, the consonants of a vocal. Parallel compression sidesteps this trade-off entirely. The dry path carries the transients untouched; the wet path carries the body, sustain, and density. Summed together, you hear both.
This is fundamentally different from using a wet/dry knob on a compressor plug-in that supports it — though that is functionally equivalent. True parallel routing gives you independent gain staging, the ability to EQ or saturate only the compressed path, and more deliberate control over the blend at every stage of the mix.
The New York Compression Technique
The New York compression technique is the most famous application of parallel compression. It was popularized by mix engineers working on high-energy rock and R&B records in Manhattan studios during the 1980s and 1990s. The term describes a very specific workflow applied primarily to drum buses, though its principles transfer to any source.
The method: create a parallel send from your drum bus to a dedicated compression channel. On that channel, apply extreme compression — a high ratio (8:1, 10:1, or even brick-wall limiting), a fast attack, a moderate-to-slow release, and a threshold set low enough to achieve 15–20 dB of gain reduction. This crushes the drum bus into a dense, sustain-heavy, almost distorted version of itself. Then blend this crushed signal back under the dry drums at a low level — typically 15 to 30 percent of the dry signal's perceived level.
What you hear is a drum track with a huge, room-filling size and incredible sustain in the toms and snare body, while the attack transients of the kick and snare remain crisp and defined from the uncompressed dry path. It is the sonic hallmark of countless classic records.
Chris Lord-Alge, Michael Brauer, and Andy Wallace are frequently cited as engineers who used heavy New York parallel compression on drum buses throughout the 1990s and 2000s. The technique became so widespread that many DAWs — including Ableton, Logic, and Pro Tools — include it as a default template in their drum bus channels.
The key variables in the New York technique are the release time on the compressed path and the blend level. A very short release (50–80ms) creates an audible pumping that can feel rhythmic and exciting in EDM or trap. A longer release (200–400ms) creates a smoother, more natural swell that works better in rock and acoustic genres. The blend level is a taste decision: a low blend (10–20%) is barely perceptible but noticeably adds weight; a higher blend (40–60%) creates a more overt, hyper-compressed aesthetic.
Settings, Signal Chain, and DAW Setup
Understanding the technique conceptually is one thing — implementing it cleanly in your DAW is another. Here is a reference table covering starting settings for different source types, followed by DAW-specific routing notes.
| Source | Ratio | Attack | Release | Blend | Compressor Type |
|---|---|---|---|---|---|
| Drum Bus | 8:1 – ∞:1 | 1–5 ms | 100–300 ms | 15–35% | VCA (1176, SSL G) |
| Snare | 10:1 – ∞:1 | 5–15 ms | 80–200 ms | 20–40% | FET (1176) |
| Bass Guitar | 4:1 – 8:1 | 10–30 ms | 150–400 ms | 30–50% | Opto (LA-2A style) |
| Lead Vocal | 4:1 – 8:1 | 15–40 ms | 200–500 ms | 20–40% | VCA or Opto |
| Mix Bus | 4:1 – 6:1 | 10–30 ms | 200–600 ms | 10–25% | VCA (SSL G style) |
In Ableton Live: Use a Return track (Cmd+Alt+T / Ctrl+Alt+T). Route your drum group to the return with the Send knob, insert a compressor on the return, and set the return fader to 0 dB. The send level controls your blend. Set the return track's pan and level independently to compensate for any gain increase.
In Logic Pro: Use a Bus send from your drum channel strip. Create a new Aux channel receiving that bus. Insert your compressor on the Aux. Use the Send level for blend control. Alternatively, many engineers in Logic use the built-in Vintage VCA compressor on the Aux for its character.
In Pro Tools: Use an Aux Input track set to receive an internal bus from your drum group. Insert the compressor there. Set the drum group's send pre-fader so that fader changes to the drum group don't affect the parallel blend ratio.
In FL Studio: Create a mixer track, route your drum channel to it as a sidechain send, insert your compressor, and blend using the mixer track fader. Use the track routing panel to ensure the dry signal remains on its original mixer track.
Parallel compression adds level to your mix bus. After setting your blend, check your output with a LUFS meter and lower either the parallel return fader or the source fader to compensate. Never adjust the blend ratio itself to manage loudness — that changes the sound, not just the level.
One nuance that trips up many producers: the compressor on the parallel path should be set with no makeup gain or minimal makeup gain. The whole point is that the heavily reduced, squashed signal contributes body and density at a relatively low level. If you apply 10–15 dB of makeup gain on the parallel compressor to match the dry level, you've effectively built a conventional serial compressor — losing the benefit entirely. Keep the compressed parallel signal quieter than the dry signal, then bring it up with the blend fader until it sounds right.
Genre Applications and Creative Uses
Parallel compression is not a single-purpose tool. Its applications shift significantly across genres, and understanding those differences will help you apply it more deliberately rather than as a default habit.
Heavy parallel compression on 808s and sample-based drum loops adds weight and sustain without dulling the snap. Use fast attack, fast release, high ratio on the parallel path. Blend aggressively at 40–60% for that characteristically thick, punchy sound.
New York drum compression is essential here. Slow release on the parallel path extends snare and tom sustain dramatically, making live drums sound massive and room-filling. Apply to the full drum bus rather than individual elements for cohesion.
Parallel vocal compression is common in pop and R&B. A slow-attack compressor on the vocal send adds body and presence without limiting expressiveness. Blend at 20–35% and use a smooth, musical compressor character like an Opto or Neve-style VCA.
Parallel compression with a short release on synth buses and drum groups creates rhythmic pumping that can function as a rhythmic element in itself. In sidechain-heavy genres this can be used deliberately to reinforce the groove.
Use parallel compression subtly — low blend, gentle ratio, slow attack and release — to add cohesion to a live room recording without losing the natural dynamics that define acoustic performance. Less is more in this context.
Running a parallel compressed version of the full mix bus can add density and perceived loudness without heavy limiting. Use a ratio of 4:1 to 6:1 with slow attack and a blend of 10–20% for transparent glue that survives the mastering stage.
Beyond drum buses and vocals, parallel compression has interesting creative uses that are less discussed. Applying parallel compression to a reverb return — compressing the reverb tail before blending it back — creates a pumping, rhythmic reverb effect that works well in electronic and experimental contexts. Similarly, parallel compression on a room mic channel can let you blend in more ambience without the level spikes that prevent it from sitting in the mix.
One key distinction worth reinforcing: parallel compression is an additive technique. You are adding compressed signal to an already-good dry signal. If the dry signal has problems — too much low-end mud, harsh mid frequencies, timing inconsistencies — parallel compression will not fix those. It will amplify them. Always address source problems in the dry chain before introducing the parallel path.
| Compression Type | Signal Path | Transient Preservation | Best Use Case |
|---|---|---|---|
| Serial (Standard) | Single, sequential | Low to moderate | Precise dynamic control, gain staging |
| Parallel | Split, then summed | High | Adding density while keeping attack |
| Sidechain | External trigger | Controlled ducking | Rhythmic pumping, space creation |
| Multiband | Frequency-split serial | Per-band variable | Frequency-specific dynamic control |
Practical Exercises
First Parallel Drum Bus
Open a session with a drum loop or live drum recording. Create a return or aux track, send your drum bus to it, and insert any compressor with ratio at 10:1, attack at 2ms, and release at 150ms. Slowly raise the send level until you hear the sustain increase without losing the snap, then find the blend level that feels most natural.
Parallel Vocal Presence
Take a vocal track that feels thin or lacks body in the mix. Set up a parallel send to an aux with a slow-attack compressor (30ms attack, 300ms release, 6:1 ratio), and blend at 25–35%. A/B between the blend engaged and bypassed while the full mix plays to hear how it fills in the low-mid body of the vocal without touching the sharp consonants.
Parallel Mix Bus Density
On a finished mix, create a parallel send from your stereo bus to an aux receiving a VCA-style bus compressor (4:1, 20ms attack, 400ms release, threshold set for 8–10dB gain reduction). EQ the parallel return to cut below 80Hz and above 10kHz — removing the low-end that muddies the sub and the top-end that fights the dry mix. Blend at 10–20% and check loudness compensation with a LUFS meter before and after, adjusting output fader to maintain the same integrated loudness.