Early reflections place a sound, the tail sets the room

The vocal that is either pasted on or washed out

A dry vocal over a finished beat sounds pasted on: stuck to the front of the speakers while everything else lives somewhere. You raise the send to the hall. Now the vocal is in a space, but the ends of words smear into the next ones and the chorus turns into a cloud. You pull the send back and it is pasted on again.

The send is moving two different things at once. A reverb has an early part, a handful of reflections in the first few tens of milliseconds, and a late part, the dense tail that dies away. They change the sound in different ways, and the send raises both by the same number of decibels.

Why it matters

One knob, two jobs

The arrive while the word is still sounding. They add to the dry sound and tell the ear about the surfaces close to the source, and the ones that come from the sides make the source itself seem wider. The tail keeps going after the word has ended. It tells the ear how big and live the whole room is, and it fills the gaps between words and hits.

Figure 1. Two reverbs that both move a vocal back, drawn as a sketch of the first 400 ms. The first puts most of its energy in strong early reflections before the 80 ms line. The second has weak early reflections and a long tail that is still sounding when the plot ends.

Both of the reverbs in the figure push a vocal away from the speakers, because both lower the balance of direct to reverberant sound, one of the main distance cues. The first does it mostly with early energy, so the voice gets a body and a place while the gaps stay fairly clean. The second does it with the tail, so you hear more room and less word. Which one you want depends on the song, but you can only choose if you can move them separately.

Many algorithmic reverbs have separate early and late level controls, or an early/late balance. If yours does not, use two returns: a small room with the decay as short as it goes for the early part, and a plate or hall for the tail.

The science

Early and late energy

Room acoustics splits a reverb at a time boundary. ISO 3382-1 defines clarity as the ratio, in decibels, of the energy arriving in the first 80 ms after the direct sound to the energy arriving after it, with a 50 ms boundary for speech:

C_{80} = 10 \log_{10} \frac{\int_0^{80\,\text{ms}} p^2(t)\,dt}{\int_{80\,\text{ms}}^{\infty} p^2(t)\,dt}

Here $p(t)$ is the sound pressure of the room's . The higher the number, the more the early energy outweighs the late.

The two sides of the boundary do different jobs. Bradley, Sato and Picard (2003) added early reflections to speech in listening tests and found that they raised the effective signal-to-noise ratio, so listeners understood more words. In their analysis of measured rooms for speech, early reflections raised the effective signal-to-noise ratio by up to 9 dB. Those were speech tests, but they are a reasonable guide for a lyric that has to get through a band. Early energy that arrives from the sides also changes how wide the source seems: Barron and Marshall (1981) linked spatial impression in concert halls to early lateral reflections. The feeling of being surrounded is a later effect. Bradley and Soulodre (1995) found that listener envelopment depends on strong lateral reflections arriving 80 ms or more after the direct sound.

Decay time decides how much of a reverb ends up late. A tail that falls 60 dB in $T$ seconds keeps $10^{-6t/T}$ of its energy after time $t$ , so for a simple exponential decay that starts with the dry sound, the share after 80 ms is:

\text{late share} = 10^{-0.48/T}

For a 1 s decay that is 33%. For a 2 s decay it is 58%. Half the energy is late at $T = 0.48 / \log_{10} 2 \approx 1.6$ s.

Figure 2. Share of a reverb's own energy that arrives more than 80 ms after the dry sound, for a simple exponential decay worked out as 10^(-0.48/T). Past a decay of about 1.6 s, more than half of the reverb is late energy. Adding 40 ms of pre-delay to the 2 s reverb raises its late share from 58% to 76%.

pushes the same tail later. With 40 ms of pre-delay, only 40 ms of the reverb fits before the boundary, so the 2 s reverb goes from 58% late to 76% late. Pre-delay keeps the start of each word clean, but it moves reverb energy out of the early window, not into it.

In the demo, try moving the decay with the level left alone, then the level with the decay left alone. Listen for whether the notes move back or the room around them grows.

Listen

Push a sound forward or back with reverb

A short melody in a generated room. Pre-delay, decay and level each change how far away the notes feel.

Loading demo…

Demo volume100%

Start with your speakers or headphones low.

Try it

Build the early part and the tail on separate returns

Loop a chorus with the full mix playing and a dry lead vocal. Remove any reverb already on the vocal.
Make return A: a room reverb, 100% wet, decay as short as it allows (around 0.3 to 0.5 s), pre-delay 0 ms. If it has early and late controls, turn the late level all the way down.
Make return B: a plate or hall, 100% wet, decay 2 s, pre-delay 20 ms, early reflections at minimum if the reverb lets you.
Send the vocal to A only. Raise the send until the vocal stops sounding pasted on, and write down the send level.
Mute A and send to B only. Raise it until the vocal sits as far back as it did with A. Compare the consonants and the gaps between lines.
Bring A back at its level and set B 6 dB below the level you found in step 5. Raise B slowly until the gaps between lines feel alive, and stop there.
Switch return A to mono, then back to stereo, and listen to the width of the voice itself.

0 of 7 steps done

Return A alone usually gets the vocal out of the speakers with every word intact. Return B alone gets there too, but you hear the hall more than the voice. Together, a little tail goes a long way.

Common mistake

Asking the tail to do the early part's job

You set a long hall low under the vocal to make it sit in a space. At a low send you hear almost none of the early part, only the tail poking out in the gaps, so you push the send up and the wash comes with it. If the vocal needs a place, give it early energy first.

The second mistake is the opposite: early reflections so loud and few that they act like single echoes. One strong reflection a few milliseconds after the dry sound combs the tone, the same effect a reflective wall has on a microphone, as the lesson on room reflections in vocal recordings explains. Keep the early level below the point where the vocal starts to sound hollow or phasey.

Takeaway

Place the sound, then size the room

I set the early part first and ask one question: does the vocal sound like it is in a place? Then I add tail only until the gaps between lines feel alive. If the tail starts covering the next line, shorten it or duck it under the vocal before turning it down; the lesson on reverb that masks the next line covers both. Early energy and tail are two faders, and a mix usually needs much less tail than the reverb preset suggests.

Sources

Show all 4

Barron, M., & Marshall, A. H. (1981). Spatial impression due to early lateral reflections in concert halls: The derivation of a physical measure. Journal of Sound and Vibration, 77(2), 211-232.
Bradley, J. S., Sato, H., & Picard, M. (2003). On the importance of early reflections for speech in rooms. Journal of the Acoustical Society of America, 113(6), 3233-3244.
Bradley, J. S., & Soulodre, G. A. (1995). Objective measures of listener envelopment. Journal of the Acoustical Society of America, 98(5), 2590-2597.
ISO 3382-1:2009. Acoustics: Measurement of room acoustic parameters, Part 1: Performance spaces. International Organization for Standardization.

Early reflections place a sound, the tail sets the room

In short

The vocal that is either pasted on or washed out

One knob, two jobs

Early and late energy

Push a sound forward or back with reverb

Build the early part and the tail on separate returns

Asking the tail to do the early part's job

Place the sound, then size the room

Check yourself

Sources

Next lesson · 27 of 44

Written by Virzy Guns

Later in Mixing & Mastering