Plate vs hall vs room vs spring reverb

At a glance

Where the four names come from

Before digital reverb arrived in the 1970s, a studio that wanted to add reverb after recording had three options. It could play the sound into a real room with hard walls, called an echo chamber, and record what came back, or it could send it to a plate or a spring tank. Rooms and halls are named after real spaces, and plates and springs after the machines.

In a plug-in, all four are now either an algorithm, meaning a network of delays built to behave like that space or machine, or an impulse response recorded from a real one. The algorithmic vs convolution reverb lesson covers both approaches. Either way the names have stuck, and they describe a sound: pick the Plate mode in Valhalla VintageVerb and you get something that behaves like a steel sheet, even though there's no steel involved.

The lab above builds all four in code. Its cards are in the same order as this page's title, so keys 1 to 4 pick plate, hall, room and spring. The impulse response screen shows what each type does to a single click, and the Right now panel explains the type or control you last touched, with numbers measured from what you're hearing. For a tour of every control, read what is reverb? first.

Room reverb

A room reverb copies a small space like a studio live room or a bedroom. Its walls are close, so the first reflections come back almost at once. Sound travels about 34 cm in a millisecond, which means a wall 3 metres away sends back a reflection that has travelled 6 metres, arriving roughly 17.5 ms after the direct sound. The other surfaces all do the same within a few milliseconds of each other.

Reflections that arrive within about 30 ms fuse with the original in your hearing. You hear them as part of the sound itself, which gets thicker and seems to come from somewhere, and that's why a room is easier to feel than to hear. In the lab's room, 14 early reflections land in the first 27 ms, and with the decay at 0.8 seconds the lab measures the tail at 785 ms, short enough to be gone before the next beat at most tempos.

That makes rooms the usual reverb for drums. A kit recorded in a good room has the room on its own microphones, and a room reverb gives programmed drums or separately recorded samples the same thing, so they start to sound like one kit played in one place. Rooms suit acoustic guitars and close vocals for the same reason, and a lot of modern pop and rap vocals use little more than a short room.

The problem with a small room is boxiness. Its reflections arrive so soon that they mix with the dry sound and cancel some frequencies while boosting others, the same comb filtering you get from a microphone too close to a wall. If a room makes a vocal sound like it's in a cupboard, pull the early reflections down or make the room bigger.

Hall reverb

A hall copies a concert hall, and its walls are a long way off. If the nearest side wall is 15 metres away, its reflection travels 30 metres and arrives about 87 ms after the direct sound, so a hall's early reflections are fewer and further apart. The lab's hall has 9 of them, spread over the first 112 ms.

The tail also takes longer to build. In a big space the echoes need time to multiply into a smooth wash, so a hall swells up for a moment before it fades. In the lab that swell takes about 280 ms, and it's a big part of why a hall sounds large even before you notice how long it lasts.

Halls built for orchestral music usually have about 2 seconds of decay, and plug-in halls go much longer. Their low end rings on longest, as it does in many real halls, because air and soft surfaces soak up treble on every bounce and leave the bass. Set the lab's hall to 2.6 seconds and the tail as a whole measures 2.85 seconds, because its lows outlast the mids that the decay control is set for.

Halls suit anything slow and sustained: strings, pads, piano, choirs and ballad vocals. They're also the obvious send when a whole arrangement should sound like it's in one big space. On fast parts the long tail fills every gap and blurs the rhythm, and all that low end can pile up into mud, so high-pass the return at 200 to 400 Hz before you blame the hall.

Plate reverb

A plate is a machine. The best known is the EMT 140, made by the German company EMT from 1957: a thin sheet of steel about 2 metres by 1, hung under tension in a frame. A driver near the centre shakes the sheet with the audio, pickups on the steel turn its vibration back into a signal, and a damping panel moved closer to or further from the plate shortens or lengthens the decay.

There's no air and no walls for the sound to bounce off on its way to a listener, so a plate has no early reflections at all. The vibration spreads out through the steel and back from its edges. In a thin plate higher frequencies travel faster than low ones, so the waves smear into a dense wash almost at once. The lab's plate tail reaches full level in about 5 ms, where the hall takes roughly 280 ms to swell up.

Plates are also bright. A room loses treble to the air and its surfaces on every bounce, but a steel sheet loses comparatively little, so a plate's highs hang on almost as long as its mids. You can see it in the figure further down, where the plate's top band lasts 83% as long as its mids and the room's lasts 55%.

That combination is why a plate is the usual first choice for a vocal. A dense start means no slapback to blur the words, a little pre-delay keeps each word clear, and the bright, smooth tail sits behind the voice instead of under it. Plates do the same for snares, where they make each hit sound longer and bigger, and for synth leads. The downside of all that top end is that it brings out S and T sounds, so roll the top off the send or put a de-esser before the reverb.

Spring reverb

A spring tank is a metal box with two or three coiled springs stretched inside. A small driver at one end shakes the springs with the signal, and a pickup at the other end turns their movement back into sound. Hammond developed it for its organs, and Fender put springs into its guitar amps in the early 1960s; plenty of guitar amps still have one.

The sound comes from the way a wave travels along the spring. It reflects off the far end and comes back, again and again, so the echoes repeat at a regular spacing, which in the lab's spring at 45% size is about every 58 ms. A spring also carries different frequencies at different speeds, so each echo arrives as a quick sweep in pitch, the chirp people describe as a boing or a drip.

Springs have a narrow range as well. They pass little deep bass and lose their top end quickly, so a spring sounds thin and a little lo-fi next to a plate. In the figure below, the spring's top band falls 60 dB in 0.68 seconds while its mids take 2.16, the biggest drop of the four. The lab's spring thins out its low end too, with a 110 Hz filter that every trip passes through.

Electric guitar is the natural home for a spring, and surf guitar is built on it. Dub engineers used springs on drums and vocals too, and knocking a real tank makes the springs crash, a sound dub records use on purpose. On ordinary drums the sharp transients make a spring clang, so filter the send or keep the level low unless that's the sound you want.

Decay at different frequencies

A decay setting is a single number, but real spaces and machines rarely fade at the same rate across the whole spectrum. The low end, the mids and the top end each have their own decay time, and the balance between them is a big part of what makes a plate sound like a plate. Plug-ins expose it with controls like damping, a bass multiplier or a decay EQ, and the four types start from very different shapes.

With every decay control at 2 seconds, the hall's lowest band rings on longest, at 2.38 seconds against 2.12 for its mids, which is why it sounds warm and full. The plate is the flattest line. The room holds its lows level with its mids, then loses its top end a little faster than the hall does, down to 1.15 seconds above 4 kHz. The spring falls off a cliff above 4 kHz, so its tail goes dark and narrow quickly.

Matching the decay times won't make two reverbs sound alike: two tails can both measure 2 seconds and still sound nothing like each other, because one is bright all the way down and the other has lost its top end within a second.

Same settings, different sound

Set all four types to the same decay, size, damping and pre-delay, and they still sound different. Try step 5 in the lab and press 1 to 4 while the vocal plays.

Listen to the first moment after each word. The hall and the room both start with early reflections, the hall's spread out and the room's tight together, while the plate and the spring have none. Then listen to the build: the plate is dense at once, the room nearly so, the hall swells and the spring stutters. Last, listen to the tone as it fades, where the plate stays bright and the spring goes dark first.

Decay is the easy thing to change afterwards, so pick the type for its start and its tone, then set the length to the tempo.

Which reverb type for which sound?

These are where to start, and plenty of good mixes break them. The flagship lesson has starting settings for each sound as well.

How to choose a reverb type

1. Decide what the reverb is forIf the part should sound like it was played in a real place, use a room or a hall. If it should just sound bigger and smoother, a plate is usually the better tool, because it has no early reflections to tell your ears how far away the walls are.
2. Listen to the gapsPlay the busiest section. Fast drums and quick vocal lines leave little space for a tail, so try a room or a short plate first; slow parts can carry a hall.
3. Listen to the start of each soundIf the reverb smears consonants or drum hits, try a plate with 20 to 40 ms of pre-delay, or a smaller room with quieter early reflections.
4. Check the tone against the mixIn a bright, busy mix a plate can add too much top end, so try a damped hall. In a dark mix a plate can add the air it's missing.
5. Compare at the same levelPresets vary a lot in level, and the louder one nearly always sounds better at first. Match them before you choose, which the lab's A/B does for you.
6. Keep to two or three spacesA short room and a plate or hall, each on a send, cover most mixes, and every part that shares a send sounds like it's in the same place. The insert vs send lesson shows the setup.

Common mistakes

A hall on everything

Halls sound impressive on a soloed part, so it's tempting to send everything to one. In a full mix the long tails and the warm low end stack up and push the whole song back. Give the drums a room and the vocal a plate, and save the hall for the parts that need to sound big.

Choosing by the name

Every plug-in's version of each type is someone's interpretation, and one company's room can be longer than another's hall. Use the names as a starting point and judge the result by ear. The lab's impulse response screen shows you what to listen for.

A bright plate on a bright vocal

A plate keeps its highs, so every S in the vocal turns into a hiss in the tail. Cut the send above 6 to 8 kHz, or de-ess the signal going into the reverb, and the plate stays smooth.

A spring on drums by accident

A guitar amp's spring sounds fine on a guitar, but put a spring plug-in on a guitar track with drum spill on it, or on a send the drums also feed, and every loud hit goes through the springs and comes out as a clang. Check what else is on the track before you add the spring.

Training your ears to tell them apart

The quiz under the lab plays a hidden room, hall or plate at matched loudness and asks which one it is, with the screens covered. The tails are different lengths, but don't rely on that. Listen to how each one starts: the plate is instantly dense and bright, the hall takes a moment to bloom, and the room is a quick, close cluster that's gone almost as soon as the sound stops. Your best score out of eight is saved in your browser.

Questions people ask

Keep going