Tutorials Audio

Reverb and Delay Explained

By Infinite Creation · Beginner · ~30 min

A 1960s mid-century comic book style editorial illustration showing two audio engineers at a mixing desk, with a mixing desk, a studio microphone on a boom arm, an audio waveform display visible in the frame.

Overview

Reverb and delay are the two main time-based effects in audio production. While compression and EQ shape the level and tone of a sound, reverb and delay place it in space: giving it a sense of room, distance, and dimension. Used subtly, they make dry recordings feel natural and present. Used more obviously, they become creative effects that define a song's character. Understanding each separately is the first step to using both intentionally.

What You Need

A DAW with a reverb and delay plugin (your DAW's built-ins are fine for learning)
A dry vocal or instrument recording to experiment on
Headphones or monitors

Steps

What reverb does

When sound occurs in a physical space, it bounces off walls, ceilings, and floors, reaching your ears as a wash of thousands of reflections that decay over time. Reverb plugins simulate this. They take a dry signal (your recording) and add a synthetic or captured room response to it, placing the sound inside a virtual acoustic space. The key numbers are pre-delay (the gap before reflections begin, simulates distance between source and walls), decay time (how long the reverb tail lasts), and wet/dry mix (how much reverb vs dry signal you hear).

Reverb types and their characters

Room reverb: small, tight, natural-sounding, mimics a treated recording room. Great for adding presence to a dry voice without it sounding like it's in a cave. Hall reverb: large, smooth, rich, mimics a concert hall. Used on orchestral strings, pads, backing vocals. Plate reverb: a metallic, dense reverb from a physical steel plate resonator. Warm and vintage, the classic snare and lead vocal reverb of the 60s to 80s. Spring reverb: bouncy, lo-fi, characteristic of guitar amplifiers. Convolution reverb: captures the impulse response of a real space (a cathedral, a bathroom, a specific studio) and applies it mathematically. The most realistic but the most CPU-intensive.

Key reverb controls

Pre-delay (10 to 30ms): Delay the reverb tail slightly so the initial transient of the sound hits cleanly before the wash begins. This keeps vocals intelligible: without pre-delay, the reverb smears into the attack and muddles the word. Decay / RT60: How long the tail lasts. Short (0.3 to 0.8s) for natural presence. Medium (1 to 2s) for character. Long (3s+) for dramatic effect. Match decay to tempo, tails that extend past the next beat cause rhythmic mush. Damping: High-frequency damping makes the reverb tail roll off at the top end, simulating soft furnishings. Reduces harshness. Wet/Dry mix: For vocals on a send channel, set the plugin to 100% wet. You control the mix with the send level on the track.

What delay does

Delay captures the input signal and plays it back after a set time, like an echo. Unlike reverb's thousands of overlapping reflections, delay is one (or a few) distinct, audible repeat(s). It adds space, width, and rhythmic interest. A single delay repeat can make a vocal feel larger without washing it out. Multiple repeats create the classic echo effect. The key control is delay time (the gap between the original and the repeat, measured in ms or synced to BPM), and feedback (how many times the repeat repeats, how quickly it fades). Use our BPM to Delay Time Calculator to find the right ms values for any tempo.

Delay types and practical settings

Slapback delay (50 to 150ms, 0 feedback): A single short echo used on vocals, especially in country and rockabilly. One quick repeat that adds size without obvious echo. No feedback so it's heard once and gone. Ping-pong delay: Alternates repeats between left and right channels, creates stereo width and movement. Keep feedback low (2 to 3 repeats) or it clutters. Tempo-synced delay: Delay time is set to a note subdivision of the song's BPM (eighth note, dotted eighth, quarter note). The repeats fall rhythmically in time with the track, a classic pop and electronic music technique. Dotted eighth delay (the Edge from U2's main technique) is the most commonly used.

When to use reverb vs delay

They solve different problems. Use reverb when you want to place a sound in an acoustic space: to make a dry vocal sound like it was recorded in a room, or to give a snare the sense of a large hall. Use delay when you want to add rhythmic interest, width, or size without washing out the sound. Many producers use both together: a short delay feeds into a reverb, so the delay repeat has its own sense of space. For podcast and VO audio: typically use neither, clean, dry audio is the professional standard for voice. If there's unwanted room sound in a recording, remove it with noise reduction rather than mask it with reverb.

Tested: how much tail is left at the next beat

We tested step 3's advice to match decay to tempo on 2026-09-25 with FFmpeg 8.1.2 and numpy. We built a reverb for each of the three Practical Settings further down this page: noise that dies away 60 dB over the decay time and starts after the pre-delay, taking the middle of each pre-delay range. We rendered a single click through each one with FFmpeg's convolution filter, afir, measured the decay from the result, and read how much of the tail was still to come when the next quarter-note beat arrived at 90, 120 and 140 BPM.

Setting from this page	Decay, measured	Tail left at the next beat, 90 BPM	Tail left at the next beat, 120 BPM	Tail left at the next beat, 140 BPM
Vocals: 1.5 s decay, 20 ms pre-delay	1.50 s	-25.8 dB	-19.3 dB	-16.4 dB
Drums: 0.5 s decay, 7.5 ms pre-delay	0.50 s	Gone, below -60 dB	-59.3 dB	-50.7 dB
Pads: 3 s decay, 90 ms pre-delay	3.01 s	-11.4 dB	-8.2 dB	-6.8 dB

Only the drum setting was gone, or nearly, by the next beat. At 120 BPM a beat lasts 500 ms, and the 0.5 second drum reverb had fallen 59.3 dB by then, close to the 60 dB that defines its decay. The vocal setting was only 19.3 dB down and the pad setting 8.2 dB. Read literally, step 3's warning about tails that extend past the next beat rules out this page's own vocal and pad settings at every tempo we tried. What decides whether a tail muddies the next note is how far down it is when the note arrives, and the table gives that.

The figures follow a simple rule, which the measurement matched to within 0.2 dB: the tail falls 60 dB over the decay time, starting after the pre-delay. So at 120 BPM with 20 ms of pre-delay, a tail that has to be gone by the next beat needs a decay of 0.48 seconds or less. One that only has to be 20 dB down can last three times as long, 1.44 seconds.

Pre-delay moves the whole tail later, so it trades a clean start for more tail at the next beat. The pad setting's 90 ms left 1.8 dB more at 120 BPM than the same reverb with no pre-delay.

We also measured a real room to set these settings against: a 3 by 2 metre office with bare walls and floor, a desk, no curtains and no other furniture. We played an 8 second sine sweep from 50 Hz to 20 kHz through the MacBook Pro's speakers, recorded it three times on a Shure MV7i microphone, and worked out the room's decay in each octave from 500 Hz up: the time to fall 60 dB, worked out from its fall between 5 and 25 dB down. It was 0.48 seconds at 500 Hz and 0.62 seconds at 1 kHz, falling to 0.43 at 4 kHz and 0.35 at 8 kHz. From 1 kHz up the three takes agreed within 0.03 seconds. So the 0.5 seconds given for a small room in Reverb Parameters below is close to what this room measured, and the 1.5 second vocal setting is about 2.7 times as long as the room itself. The room also died away fastest at the top, which is what the damping control in step 3 imitates. Below 500 Hz the laptop speakers put out too little for the takes to agree, so we leave those bands out.

Step 2 calls convolution the most CPU-intensive kind of reverb. It did cost more than the one other effect we could run, though little in absolute terms on this Apple M1 Max. A minute of stereo audio through one 3 second impulse response took 0.17 seconds of processor time at FFmpeg's default partition size and 0.64 seconds at 64-sample partitions, which add less latency. That is about 1.1 percent of one core. The 100 ms slapback delay took 0.03 seconds, and reading the file with no effect 0.01 seconds.

The command below needed two changes from FFmpeg's defaults, both found by running it. On its own, afir stops when its input stops, which cut the tail off: 4 seconds in, 4 seconds out. apad adds silence for the tail to ring into. And with its default scaling, a 3 second impulse response came out 42 dB quieter than the input. With irnorm=2 the output measured within 0.5 dB of the input. The result is all reverb, what a send returns, to be mixed under the dry track as step 3 describes.

To try a convolution reverb from an impulse response file: ffmpeg -i vocal.wav -i room-ir.wav -filter_complex "[0:a]apad=pad_dur=3[in];[in][1:a]afir=irnorm=2" vocal-reverb.wav

Validated. Measured 2026-09-25 with FFmpeg 8.1.2 and numpy on an Apple M1 Max Mac running macOS 26.6.2. The reverb is a model: decaying noise with no early reflections and the same decay at every frequency. A plugin's decay setting may not match a measured decay exactly, and processor time depends on the machine and the plugin. We did not test any reverb or delay plugin named on this page, or an algorithmic reverb. The room figures come from one room, one speaker and one microphone position, and we did not record the microphone's distance or angle. We could not read the microphone's own processing settings, but a fixed EQ in it would barely matter here: 6 dB of EQ applied to the recordings moved the room figures by at most 0.03 seconds. The recordings are kept with the script that analyses them.

Pro Tips

Always apply reverb and delay on a send/return (aux) channel, not directly on the track insert. This lets you control the wet mix with the send level and use one reverb instance for multiple tracks, saving CPU.
High-pass filter your reverb return at around 100 to 150Hz. Bass frequencies in reverb tails create mud fast, cut them and your mix will be cleaner immediately.
Automate reverb sends, increase the wet level on the last word of a phrase, lower it at the start of a new phrase. This is a standard pop production technique for vocal presence without constant wash.
Free options: Valhalla Supermassive (reverb/delay, free), OldSkoolVerb (free plate), TAL-Reverb (free algorithmic). All are high-quality and used by professionals.

Knowledge Base

What You'll Learn

Use reverb and delay to add space and depth to your mix. Understand decay time, pre-delay, and when to use each effect. Create professional-sounding recordings with proper space.

Reverb vs. Delay

Reverb: Simulates room acoustics. Many echoes so densely packed they're not distinguishable. Ambient, spacious feel.
Delay: Distinct repeats. Noticeable echoes, musical effect. Synced to tempo creates rhythmic repeats.

Reverb Parameters

Decay Time (RT60): How long sound takes to drop 60 dB. Small room: 0.5s. Large hall: 2-3s.
Pre-delay: Time before reverb starts (20-80ms). Longer = more spacious. Shorter = intimate.
Wet/Dry: Balance of original signal (dry) to reverb (wet). Usually 20-40% wet.
Early Reflections: Initial bounces. Define room character.

Practical Settings

Vocals: Medium reverb (1.5s decay), 20ms pre-delay, 30% wet.
Drums: Short reverb (0.5s decay), 5-10ms pre-delay, 15-20% wet. Tight and punchy.
Ambience/Pads: Long reverb (3+ seconds), 80-100ms pre-delay, 50%+ wet.

Delay Sync

Sync delay time to your BPM using the BPM calculator (elsewhere on this site). Quarter-note delay is most common. Dotted-eighth is popular for creating space.

Where This Fits

Reverb and delay are sends, and Mixing a Full Song covers the bus structure they return into.

FAQ

Q: How much reverb is too much?
A: If you can't understand lyrics, too much. If it sounds dry, add more. 20-30% wet is usually right.