Reverb and Delay: How Space and Time Effects Work

A voice recorded in a treated room is dry, close, almost unnatural. In real life every sound comes with reflections that tell us where we are: a cathedral, a tiled bathroom, a living room. Reverb and delay are the tools that recreate this space in a controlled way. Used well they add depth; used badly they push things back and blur them.

Delay: a copy in time

Delay is the simplest of the time-based effects: it plays back a copy of the signal after a certain interval. The essential parameters are:

• Time: the interval between the sound and its repeat, in milliseconds or in musical note values synced to the song's tempo.
• Feedback: how much of the output is sent back to the input. With zero feedback you get a single repeat; raising it produces multiple repeats that fade away.
• Mix: the balance between the original signal and the repeats.
• Filters: many delays attenuate the lows and highs of the repeats, so they stay in the background.

Times and perception

To calculate a delay in time with the song: 60,000 divided by the BPM gives the length of a quarter note in milliseconds. At 120 BPM, a quarter note lasts 500 ms and an eighth note 250 ms.

Reverb: thousands of reflections

Reverb is the sum of a great many reflections so closely spaced that they are not perceived individually. It has three phases:

1. Direct sound: arrives first.
2. Early reflections: the first ones, distinguishable by the auditory system, which convey information about the size and shape of the space.
3. Tail: the diffuse mass of reflections that gradually decays.

The main parameters

• Pre-delay: the time between the direct sound and the start of the reverb. A pre-delay of 20-40 ms separates the voice from the tail, keeping it clear even with long reverbs.
• Decay (or RT60): how long the tail takes to drop by 60 dB. A few tenths of a second for a room, several seconds for a church.
• Size: changes the density and timing of the early reflections.
• Damping: how quickly the highs fade in the tail, as happens in spaces with soft surfaces.
• Output filters: cutting the lows of the reverb keeps the mix from getting muddy.

Algorithmic or convolution

Algorithmic reverbs generate reflections with a network of delays and filters. They are highly adjustable, light on the CPU, and can create spaces that don't exist.

Convolution reverbs use an impulse response, a recording of how a real space responds to a very short sound. The plug-in "applies" that space to the signal. The result is very realistic, but less flexible and heavier to compute.

Insert or send

There are two ways to add a space effect:

• Insert: the effect sits on the track itself and is adjusted with the mix control. Simple, but every track would have its own reverb.
• Aux send: you create an effects track with the reverb set to 100% wet, and each track sends an adjustable amount of signal to it.

The send is almost always preferable: a single shared reverb gives the impression that all instruments are in the same space, uses fewer resources and lets you EQ or compress the reverb separately.

Reverb on speech and podcasts

On spoken voice reverb should be used with great caution. A dry voice sounds closer and more intelligible, especially on headphones. Two cases where it makes sense:

• Matching voices recorded in different rooms, with a very short, low-level room reverb that is almost imperceptible.
• Narrative segments or sound design, where the space is part of the story.

In my work, if a listener notices the reverb on a spoken voice, it usually means there is too much.

In short

• Delay repeats the signal after a set time; feedback and filters decide how the repeats fade.
• Reverb is made of early reflections and a tail; pre-delay keeps the voice clear.
• Algorithmic reverbs are flexible, convolution reverbs reproduce real spaces.
• Aux sends create a coherent space and let you process the effect separately.
• On spoken voice, reverb should be used very sparingly.