Are the rooms recordings of real places?

No, they are synthesized from an early-reflection pattern plus a decaying noise tail. That keeps the tool inside a browser with no sample library, at the cost of not being able to reproduce a particular venue. The result is a plausible room rather than a specific one.

Why is the output longer than the input?

Because convolution extends the signal by the length of the impulse response: the tail keeps ringing after the source has stopped. Cutting the result back to the input length would throw that tail away and produce a reverb that stopped abruptly at the end of the file.

What does the room gain figure do?

It keeps the wet signal at a sensible level as the space grows. A larger room spreads the same energy over a longer time, so without an adjustment the cathedral would sound quieter than the small room at the same mix setting.

Why is the cathedral slower to render?

Because convolution cost is proportional to the length of the impulse response. A four and a half second response takes several times the arithmetic of a four tenths of a second one, which is the price of the realism rather than a sign of a problem.

Can I use my own impulse response?

Not in this tool, which relies on its synthesized rooms. To work with a real response, a sweep would have to be recorded and processed first, which is a separate job from applying a reverb and is not part of this pipeline.

Convolution Reverb

You Might Also Need

Reverb Simulator

Add realistic room reverberation

Hallway / Room Simulator

Simulate various room acoustics

Reverb Time Calculator

Calculate reverb pre-delay and decay

Echo / Delay Generator

Create echo and delay effects

Convolution Reverb smears a recording with a room's impulse response rather than approximating one with filters. The rooms are synthesized rather than sampled, so no files have to be loaded.

What is Convolution Reverb?

Choose a file, pick one of four rooms, set the mix, and apply; the output includes the reverb tail running past the end of the source.

The controls and the rooms:

Four room buttons: a small room with a decay around four tenths of a second, a club at about a second, a concert hall at two and a fifth, and a cathedral at four and a half.
A mix from nothing to everything, defaulting to thirty five percent, balancing the dry signal against the convolved one.
Each room carries its own gain figure, from a third for the small room up to six tenths for the cathedral, which is what keeps the wet signal at a sensible level as the space grows.
The report names the room, states that the impulse response is synthetic, gives its decay figure, and repeats the mix.

The method is true convolution, which means the audio is multiplied and summed against an impulse response sample by sample — effectively the same operation as playing the recording into the room and capturing what comes back. The response here is built rather than recorded: an early-reflection pattern of five taps at twelve, twenty-three, thirty-seven, fifty-three and seventy-one milliseconds supplies the discrete first echoes that the ear reads as room size, and a decaying noise tail supplies the diffuse part that follows. The decay figure is the time for that tail to fall sixty decibels, which is the standard way a room's length is described and is the figure the room table calls rt60. Convolution has one property that matters in practice and that the implementation had to respect: the output is longer than the input by the length of the impulse response, because the tail keeps ringing after the source stops. Truncating the result to the input's length would discard the tail entirely and leave a reverb that appeared to stop dead at the end of the file, which is what the mixing loop here deliberately avoids by carrying the wet signal through to the end of the convolved buffer. The other consequence of convolution is its cost: the arithmetic is proportional to the length of the impulse response, so the cathedral is several times more work than the small room, which is why a long tail takes noticeably longer to render.

Where it differs from a sampled reverb:

The impulse responses are synthesized, so no specific real venue can be reproduced.
No impulse response file can be loaded.
Four rooms only, with no control of decay or of the early-reflection pattern.
No pre-delay or tone shaping of the wet signal.
Rendering is slower than an algorithmic reverb because the arithmetic scales with the response length.
Output is sixteen bit WAV, with the tail included.

Use it when the spatial character matters more than the parameters: placing a dry recording in a large believable space, adding a long cathedral tail to a sustained part, or giving a small room a tight but realistic ambience. When the decay should be dialable rather than chosen from a list, the reverb simulator gives direct control of feedback and damping, and when only the early reflections that establish room size are wanted, the hallway room simulator isolates exactly that cue.

How to use Convolution Reverb

Drop an audio file onto the panel and let it decode.
Choose one of the four rooms by its character rather than by its figures.
Set the mix, using a lower figure for a wet but undistracting result.
Apply the reverb and download the WAV, which carries the tail past the end of the source.

When to use Convolution Reverb vs related tools

Convolution Reverb is the tool when a realistic space matters more than adjustable parameters. Reverb Simulator gives direct control of decay and damping instead, and Hallway Room Simulator isolates the early reflections that signal room size without adding a tail. When the decay needs to fit a tempo, Reverb Time Calculator provides the figures to aim at.

Privacy & Security

This tool runs entirely in your browser — no data ever leaves your device. There is no server round-trip, no upload, no logging, and no account required. Your input is processed locally using client-side JavaScript and is never stored, transmitted, or accessible to anyone else. When you close the tab, everything disappears.