Add Reverb & Echo to Audio

Reverb puts a sound in a place. Echo puts it in time. Used together they are how a dry recording becomes a record, and the reason they belong on one page is that they interact: a 375 ms repeat inside a small room is rhythmic, and the same repeat inside a cathedral is mud. This tool shows both on one timeline so you are deciding the relationship rather than two unrelated settings.

A reverb fills the gaps between notes; a delay creates new notes in the gaps. Alone, a heavy reverb smears a mix and a bare delay sounds mechanical. Together, a short reverb behind a rhythmic delay gives depth without losing the edges — which is why almost every vocal you have heard has both, usually in small amounts.

• Vocals. A plate reverb at 20% with one soft repeat is the standard pop treatment, and the Vocal Plate preset is exactly that.
• Guitars and synths. A dotted eighth delay locked to the tempo turns a simple part into a pattern; add a hall behind it for width.
• Drums. A tight bright chamber and a sixteenth repeat keep the groove and add size without washing out the transients.
• Podcasts and voiceover. A studio room at 10% wet makes a dry close-mic recording sound like it was made somewhere, and nothing else.
• Dub and remix. Long delay, high feedback and heavy damping, all fed into a big hall — the Dub Hall preset.

Set a delay in milliseconds and you are doing arithmetic in your head: 60 divided by the tempo gives a quarter note, and everything else is a fraction of that. Every other online delay leaves you to it; one of them writes the sum out in its help text. This tool finds the tempo from your own file and offers the note instead.

The detection runs on your device on the first thirty seconds of audio, and it declines rather than guessing: speech, a single held note and silence have no tempo, and a number offered for them would be indistinguishable from a real one. When it declines you can tap the tempo in or type it. A division that would fall outside the delay's range — a half note at a very slow tempo — is marked unavailable rather than quietly clamped to something else.

• 1/4. One beat. The most obvious repeat, and the safest on anything with a strong pulse.
• 1/8 dotted. Three-quarters of a beat. The classic delay setting — it lands between beats and drives the track forward.
• 1/8 and 1/16. Half and a quarter of a beat. Tighter, more rhythmic, good on percussion and short syllables.
• Triplets. Two-thirds and one-third of a beat, for a rolling feel against a straight rhythm.

Most effects tools show you the control you moved and call that the answer. This one sends a test impulse through the identical processing chain after every run and reads what came out: how long the tail actually decays, where each repeat actually landed, and how loud the finished file actually peaks. Those numbers appear beside the file.

It matters more than it sounds. The page this replaces had a wet control wired backwards — raising it made the effect quieter — and nothing on screen could have shown that, because everything on screen came from the control itself. A measurement taken from the output is the only kind that can disagree with the settings, which is the only kind worth showing.

It makes no difference, and that is a fact rather than an opinion. Both effects are linear, and linear effects can be applied in either order for the same result. We ran a file through both orders and compared them sample by sample: the largest difference was 0.00002% of the peak, which is rounding in the last decimal place rather than an audible change.

There is one real exception. Ping-pong alternates the repeats between the left and right channels, and a stereo reverb treats the two channels differently by design, so those two do interact — measured, the two orders differ by 4.0% there. That single case is why the tool has a fixed order rather than a choice: the order that measurement preferred is the one it uses.

Five of the spaces are modelled from real dimensions, so the pattern of early reflections comes from the size of the room rather than from a list of numbers. Two are not rooms at all.

• Studio. A small treated space, about five metres. Short, dry, and almost invisible — the one to use when you want presence rather than an effect.
• Room. Eight metres, a little live. The default, and the safest starting point on almost anything.
• Chamber. Fourteen metres of hard surfaces. Dense and bright, historically a real tiled room with a speaker and a microphone in it.
• Hall. Twenty-four metres. Wide, slow to build, and the natural home for anything that needs to sound big.
• Cathedral. Forty-five metres. Very long, very diffuse, and easy to overdo — keep the mix low.
• Plate. Not a room: a sheet of steel. No discrete early reflections at all, which is why it sits behind a voice so well.
• Spring. A dispersive delay line, the sound inside a guitar amplifier. Boingy on purpose.

• Decay. How long the room takes to fall silent, from 0.2 to 10 seconds. This is the number that is measured on the finished file.
• Reverb mix. How much of the room you hear against the original. Under 30% for anything with words in it.
• Delay time. The gap before the first repeat, either in milliseconds or as a note division at your track's tempo.
• Feedback. How much of each repeat feeds the next. It sets the repeat count, which the rail shows before you run.
• Echo mix. How loud the whole set of repeats is against the original — independent of how many there are.
• Echo damping. How much darker each repeat gets than the one before it. The tape-delay character.
• Ping-pong. Alternates the repeats between left and right instead of keeping them centred.
• Pre-delay. A gap before the room answers, up to 200 ms. It keeps the original clear before the reverb starts.
• Early reflections and diffusion. How much of the first bounce you hear, and how quickly the tail becomes a wash rather than distinct echoes.
• Damping and tone. How fast the high frequencies decay compared with the low ones — the difference between a carpeted room and a tiled one.
• Stereo width. How different the left and right sides of the tail are. At zero the reverb sits in the middle.
• Tail. Ring out past the end of the source, or trim to the original length.
• Clip guard. Holds the finished file under a ceiling you set. The peak it actually reached is shown beside the file.

• Start from a preset. Eight of them cover the usual jobs. Every control stays in front of you afterwards — a preset is a starting point, not a mode.
• Set the delay before the reverb. The repeats are what you will notice first. Get them in time, then decide how big the room around them should be.
• Keep both mixes low on speech. 10% reverb and no echo is enough to stop a close-mic recording sounding airless.
• Use pre-delay to keep words clear. Fifteen to twenty-five milliseconds of gap lets a syllable land before the room answers it.
• Watch the tail figure. The rail shows how much silence the effect will add. A five-second clip with a nine-second tail is mostly tail.
• Check the measured numbers. If the measured decay and the requested decay disagree, that is the tool telling you something rather than a rounding error.

MP3, WAV, OGG, OPUS, M4A, M4B, AAC, FLAC, MKA, WebM, AIFF, WMA and 3GP are accepted, as is the audio track of an MP4, MOV or MKV — dropping a video and keeping only its sound is a normal thing to want. The output can be MP3, M4A, AAC, OGG, FLAC or WAV, or the source's own container. The list is enforced at the file picker rather than discovered halfway through a run: an unsupported file is refused immediately with a link to the converter.

Every part of this runs on your device — the reverb, the delay, the tempo detection and the measurements. Your audio is never sent anywhere, which is also why there is no queue and no file-size ceiling imposed by a server: the limit is what your own machine can hold.

Powered by optimised audio processing algorithms.

• Audio reverb
• Audio echo
• Audio equalizer
• Audio normalizer
• Audio compressor
• Audio converter