Reverb is the pattern of sound reflections that lingers in a space after the original source stops, and the delay times the page uses are what separate a usable sketch from a guess. In this tool, three fixed reflections at 29.7, 37.1, and 41.1 milliseconds do the entire reverb job — no impulse response, no diffusion network, no convolution pass. "Adding reverb to audio online" means running that disclosed effect inside a browser tab that mixes a delayed wet copy into a dry recording and exports a new WAV. The phrase describes a workflow rather than a single piece of hardware: you point a browser page at an audio file, pick an effect from six choices, set a percentage amount, and download the processed result. Nothing about the phrase implies a server render, a sign-up, or a hidden AI transformation; the most honest reading is a deterministic local operation that you can run on a laptop, repeat tomorrow, and reproduce byte-for-byte. Once those two ideas — the acoustic effect and the local browser delivery — are separated, the rest of the workflow is a matter of choosing a file, a control, and an output format.
What "Reverb" Means in Audio
Reverb begins life as a physical phenomenon. When you clap in a small bedroom, you hear the original snap, then a quick blur of reflected energy bouncing off the walls, floor, and ceiling, and then silence. In a cathedral, the same clap produces a long, slowly decaying tail because the room is larger and the surfaces are harder. Audio engineers borrow the word to describe any processing that imitates that blur, whether the simulation is rough or detailed. Dry audio is the signal before reverberation is added; wet audio is the version with reflections mixed back in; the blend ratio is often called the wet/dry mix.
The phrase "add reverb to audio" therefore means mixing a reverberant copy of a signal into the original, usually under a single percentage control. Higher percentages sit closer to a fully wet sound, where the dry source is barely audible; lower percentages feel like a small amount of room added to a clean recording. A reverbed vocal sounds farther away from the microphone than a dry one, even when nothing about the microphone position has changed.
What "Add Reverb to Audio Online" Means in Practice
Adding the word "online" shifts the question from the effect itself to the delivery method. An online reverb tool is one that runs in a browser tab rather than as a plug-in inside a digital audio workstation. To make that possible, three things must happen inside the page: the file is read from your device, decoded into a PCM buffer the browser can manipulate, processed by a JavaScript or WebAssembly effect, and finally written back out as a downloadable audio file. The "online" part is shorthand for "no installation, no plug-in chain, no project session."
In practice, what "add reverb to audio online" means to a typical reader is closer to a small, repeatable task: open a page, drop in a recording, choose reverb, set the strength, and download the result. Some pages route the audio to a server, others do the work locally using the browser's built-in audio graph. The contract matters because a server route uploads the audio and may keep a copy, while a local route never leaves the tab. Audio Effects Online is explicit about the second route: every decode, effect, and WAV export happens in the current tab (per the W3C Web Audio API 1.1 specification the browser implements), your file is not uploaded, and there is no account to create.
How a Local Lightweight Reverb Is Built
A fully modelled reverb simulates the physics of a room: thousands of early reflections, frequency-dependent decay, stereo diffusion, and a long, smooth tail. A lightweight reverb is something simpler. It stacks a small number of delayed copies of the dry signal at fixed offsets and lets the wet/dry control scale how loud each copy is. The result is less realistic than a convolution reverb but is easier to predict, easier to test, and cheaper to run in real time.
The reverb in Audio Effects Online is a three-tap early-reflection design with disclosed coefficients. The page adds copies of the dry signal at 29.7, 37.1, and 41.1 milliseconds, with weights of 0.50, 0.35, and 0.25 multiplied by the selected amount. At a 100% amount the three copies are mixed in at those full weights; at a 50% amount each tap is scaled by 0.5; at 0% the reverb contributes nothing. Those three delay times are short enough to read as "early reflections" rather than as a long hall tail, and they sit close together so the combined wet signal feels like a single quick bloom rather than a sequence of distinct repeats. That arithmetic is the entire reverb engine — there is no impulse response, no diffusion network, and no convolution pass. The design is useful for sketching a vocal or drum hit that needs a touch of space, but it deliberately does not claim to model a real room.
The complete tail is reserved before processing starts, so the last early reflection is never truncated partway through its decay. If the resulting file would push past the page's output budget of thirty million channel samples, the run is rejected and no file is produced — the page will not silently cut off the reverb. Finite samples are also clamped to the legal −1 to +1 range before being quantised to sixteen-bit PCM, which prevents the kind of digital clipping that happens when multiple delayed copies stack at full level.
Why Those Three Delay Times Matter
Delay times in a reverb are not interchangeable. Pick three delays that are too close together, under about five milliseconds apart, and the wet signal sounds like a flanger or a comb filter — a metallic sweep that draws attention to itself. Spread the same three taps across a wider window, say fifty or eighty milliseconds apart, and the result starts to read as discrete repeats rather than as room ambience. The 29.7, 37.1, and 41.1 millisecond values land in the early-reflection zone, between roughly twenty and fifty milliseconds after the dry signal, where the brain still interprets the wet copy as part of the same acoustic event. Pair that placement with decreasing weights of 0.50, 0.35, and 0.25, and the wet signal fades gently inside the same window, giving the impression of a small, hard room without producing audible repeats.
The weights also play a deliberate role. The first tap at 29.7 ms is the loudest at 0.50, so it carries most of the perceived space. The second tap at 37.1 ms sits at 0.35, reinforcing the bloom without competing with the first. The third tap at 41.1 ms at 0.25 is just enough to thicken the back of the envelope. Because every tap is multiplied by the selected amount before mixing, lowering the amount does not just turn the reverb down — it proportionally thins the entire wet envelope, which is why a 30% setting feels like a small room and a 100% setting feels like a touch more of the same room.
How to Add Reverb to Audio Online with Audio Effects Online
- Open Audio Effects Online in a current desktop browser such as Chrome, Firefox, Safari, or Edge.
- Choose one browser-decodable audio file no larger than fifty mebibytes. Accepted filename extensions include MP3, WAV, M4A, AAC, Ogg, WebM, and FLAC, though a recognised extension does not guarantee that a particular encoded file will decode in your browser.
- Pick "lightweight reverb" from the effect selector.
- Set a whole-number amount between zero and one hundred percent. Zero percent returns the dry signal essentially unchanged, modulo the floating-to-PCM rounding at the end; one hundred percent applies the full three-tap mix; intermediate values scale the disclosed tap weights linearly.
- Run the effect locally. The page decodes your file with the browser's Web Audio implementation, applies the reverb in place, and reports channel count, sample rate, input and output frame counts, output peak, and the exact WAV byte size.
- Download the processed file as a new PCM16 WAV. The original file on your disk is not modified and no copy is uploaded.
- To deepen the sound, increase the amount and rerun, or download the first result and feed that WAV back in as a new input for another effect pass.
If decoding fails because the codec or profile inside your file is not supported by your browser, the page produces no output rather than a broken WAV. If you want a guided walk-through of the same steps with notes on what to expect from each control, how to add reverb to audio in your browser breaks the same flow into plain English.
The Six Effects Available in Audio Effects Online
| Effect | What it does to the audio | Typical use |
|---|---|---|
| Echo | Adds a single 200-millisecond half-strength copy of each sample, with the full tail allocated in advance. | Slap-back vocals, rhythmic repeats, dub-style effects. |
| Lightweight reverb | Adds three early reflections at 29.7, 37.1, and 41.1 milliseconds with weights of 0.50, 0.35, and 0.25 multiplied by amount. | Quick space and depth on dry recordings, sketch-level sound design. |
| Reverse | Reads the decoded frames in the opposite direction and blends with the original by amount. | Backwards vocals, transition effects, sound-design layers. |
| Fade in | Scales from the dry level at the start of the file up to full level at the end. | Smoothly introducing a track, podcast, or sample. |
| Fade out | Scales from full level at the beginning of the file down to the dry proportion at the end. | Ending a track, ring-out effect, podcast close. |
| Normalization | Scales the largest absolute decoded sample so the peak sits at 0.95. | Maximum level without exceeding the legal PCM range; not loudness matching. |
Each effect runs in one pass only. The page applies exactly one selected transformation per run and exports one WAV. To stack effects, download the first output, reopen the file as a new input, and choose the next effect.
Reading the Amount Control and the Output Report
The amount slider is the only knob the reverb has. At zero percent the tool returns the dry signal essentially unchanged, modulo the floating-to-PCM rounding at the export stage. At one hundred percent the dry signal is mixed with the three reverb taps at their full disclosed weights. Values in between scale each tap weight linearly, which is why the page can describe the operation as deterministic: the same input file and the same amount always produce the same byte sequence, modulo the final PCM quantisation step.
After processing, the page reports the channel count, sample rate, input frame count, output frame count, output peak, and the exact WAV byte size. Output frame count is slightly larger than the input for echo and reverb because of their allocated tails; if that growth would push past thirty million channel samples, the page refuses to create a file. Output peak tells you whether the result clipped at any sample, which can happen when a high-amount delay overlaps with a loud dry section. If clipping is a problem, lower the amount, run normalization on the result, or trim the source file with Audio Cutter and try again.
What This Reverb Does Not Model
Setting the right expectation matters as much as knowing the controls. The reverb here is a disclosed three-tap early-reflection design, not a measured concert hall, plate, spring, studio chamber, or branded processor. It does not model room dimensions, frequency-dependent decay, stereo diffusion, damping, pre-delay, or a professional reverberation tail. If you need a cathedral-tail vocal or a snare that sits inside a real room, a convolution reverb loaded with an impulse response is the appropriate tool — this page is not.
Normalization, by the same logic, is not loudness matching. It finds the largest absolute sample and scales the signal so that the peak sits at 0.95 of full scale. It does not measure LUFS, perceived loudness, broadcast compliance, dynamics, true peak between samples, or inter-file equal loudness. A silent file stays silent because there is no peak to scale toward. If your goal is broadcast-ready levels, normalize first and then apply loudness matching in a separate step with a different tool.