Reverb can be added to a video's extracted audio track using a browser-based audio effect tool that decodes a local audio file, applies a disclosed three-tap early-reflection effect, and exports the result as a standard PCM16 WAV file without uploading anything to a server. The workflow treats the video's audio as the audio track lifted from a clip, exported or saved as an audio-only file (commonly M4A or WAV), and then run through the effect locally in the current tab. Because every step happens in the browser, the source file never leaves the user's machine, which keeps large project audio private and avoids account sign-up, watermarks, or render queues.
Why Add Reverb to a Video's Audio Track
Reverb on a video soundtrack is most often wanted when the original recording sounds dry, when dialogue was captured close to the microphone, or when ambient music needs more space before it sits under a scene. Creators who edit travel clips, interview footage, or short social videos frequently want the vocals to feel "inside" a room rather than floating in dead air. The same problem shows up in voice-over work, podcast video, and tutorial recordings where a flat narration track can feel harsh against outdoor b-roll.
Doing this online usually means opening a browser tab, pointing it at an audio file, choosing reverb, and downloading the processed result. The catch with most cloud services is that they require uploading the file, which adds time, can hit a size cap, and raises privacy questions for unfinished projects. A local browser tool skips the upload entirely and still gives a usable reverb draft for storyboarding, rough cuts, and demo reels.
How the Local Effect Tool Works
Audio Effects Online is built around six fixed offline operations: a single 200-millisecond echo, a lightweight three-tap reverb, reverse, fade in, fade out, and peak normalization. Each operation is deterministic, which means the same input and the same amount setting always produce the same output samples. The tool does not call any audio model, convolution engine, or AI processor; it simply reads the decoded sample frames, blends or adds them according to a published formula, and re-encodes a new WAV.
Reverb specifically is a disclosed early-reflection design rather than a room simulator. Copies of the dry signal are added at 29.7, 37.1, and 41.1 milliseconds with weights of 0.50, 0.35, and 0.25, each multiplied by the chosen amount. At 100% amount the three taps are added at their full base weights; at lower amounts the weights scale linearly downward, and at 0% the reverb contributes nothing and the output equals the dry signal. The result adds a deterministic space or repetition cue that sits on top of the dry signal, but it does not model room dimensions, frequency-dependent decay, stereo diffusion, damping, pre-delay, or convolution with an impulse response. Because the reverb is short and predictable, it works well as a sound-design draft, but it is not a measurement of any real hall, plate, spring, studio, or branded processor.
The complete tool description and bounded behavior are available on the Audio Effects Online page, where the effect choices, amount range, output format, and budget limits are listed.
The Six Available Effects
| Effect | What it does |
|---|---|
| Echo | Adds a single 200 ms delayed copy at half strength, scaled linearly by the amount. |
| Lightweight reverb | Adds three reflection taps at 29.7, 37.1, and 41.1 ms with weights 0.50, 0.35, and 0.25, scaled by the amount. |
| Reverse | Reads sample frames in opposite order and blends the result with the original by the amount. |
| Fade in | Linear ramp from the dry proportion at the start to full level at the end. |
| Fade out | Inverse linear ramp from full level at the start to the dry proportion at the end. |
| Normalization | Finds the largest absolute decoded sample and scales the file toward a 0.95 sample peak. |
Add Reverb to a Video's Extracted Audio Step by Step
- Extract the audio track from the video first. Save the clip's audio as M4A, WAV, or another browser-decodable file. If the source is a long clip and only a section is needed, an audio cutter built for this step is documented in how to cut audio from video and export a WAV; the resulting WAV can be used directly as the input for the next step.
- Open Audio Effects Online and choose the extracted audio file. The file must be browser-decodable and at most 50 MiB. A recognized name such as MP3, WAV, M4A, AAC, Ogg, WebM, or FLAC is accepted, but actual codec support depends on the browser and operating system, so a recognizable extension does not guarantee successful decoding.
- Select "lightweight reverb" from the effect list and set an amount between 0% and 100%. The amount is a whole-number percentage that scales the three reverb taps linearly. A first attempt at 35% to 50% usually reveals how dense the reflections sound without overwhelming the original recording.
- Apply the effect locally. The browser decodes the file with its Web Audio implementation, allocates the complete reverb tail up front, processes every sample, and reports the channel count, sample rate, input and output frame counts, output peak, and exact WAV byte size.
- Inspect the reported numbers and preview the result. A peak below 1.0 means no clipping occurred; a peak sitting at 1.0 means dry and delayed samples overlapped and the legal -1 to +1 range capped the output.
- Download the new PCM16 WAV. The file is a fresh interleaved little-endian RIFF/WAVE. It is a separate processed copy; the untouched original stays on disk so the work is reversible.
What the Reverb Actually Does to the Audio
The three reverb taps are short and fixed, which keeps the effect predictable and fast but also means the result is not a realistic room. The table below lists the exact published coefficients the tool uses for every reverb run.
| Tap | Delay | Base weight (multiplied by amount) |
|---|---|---|
| 1 | 29.7 ms | 0.50 |
| 2 | 37.1 ms | 0.35 |
| 3 | 41.1 ms | 0.25 |
At 100% amount, each tap is added at its full base weight; at 50% the weights halve; at 0% the reverb contributes nothing and the output equals the dry signal. Because the three taps sit within roughly 12 ms of each other, the effect thickens the source rather than producing a long, lingering tail. The result has audible space, but it does not simulate room dimensions, frequency-dependent decay, stereo diffusion, damping, pre-delay, or convolution with an impulse response. Anyone needing that level of realism should reach for a dedicated convolution reverb plugin after this draft step.
Input Limits, Decoding, and the WAV Output
Every run has to fit inside three independent budgets: a 50 MiB encoded-file cap, a 30 million channel-sample limit on the decoded input, and a 30 million channel-sample limit on the complete effect output. Reverb and echo can pass the input budget yet fail the output budget because their tails add frames; when that happens the tool reports an error and produces no file, rather than truncating the tail. The decoded audio must also fall inside 1 to 8 equal-length channels at 8,000 to 192,000 Hz, with an integer amount from 0 through 100. Inputs outside those windows are rejected up front.
The output is always a new interleaved little-endian PCM16 RIFF/WAVE. According to Microsoft's WAVEFORMATEX structure, PCM block alignment is channels multiplied by bits per sample divided by eight, which at 16 bits produces 2 bytes per mono frame and 4 bytes per stereo frame. The W3C Web Audio API 1.1 specification defines how the browser decodes an AudioBuffer and how its length is measured in sample frames, which the tool uses to verify the budgets.
The downloaded file preserves the processed channel samples and sample rate but does not preserve the source codec, bitrate, compression quality, ID3 tags, artwork, chapters, cue points, loop markers, or any container metadata. Anyone who needs those artifacts must keep the original file on disk.
Build a Chain and Avoid Common Pitfalls
The tool applies exactly one effect per run, so combining reverb with fade-in or normalization requires two passes. The first pass runs reverb and saves a new WAV; the second pass opens that WAV and runs fade or normalization on the result. Repeated lossy decode and PCM processing can reveal source artifacts, and stacking normalization twice will not make a file louder the second time because the peak has already been pulled toward 0.95.
Strong reverb amounts can clip where dry and delayed samples overlap. The tool constrains finite samples to the legal -1 through +1 range before WAV quantization, which prevents wraparound but also caps how loud the result can be. If the reported output peak sits at 1.0 after a heavy reverb, lowering the amount or running peak normalization as a second pass brings the level back into a usable range.
Normalization here scales toward a 0.95 sample peak. It is not loudness matching: it does not measure LUFS, perceived volume, broadcast compliance, dynamics, true peak between samples, or equal loudness across different recordings. Silent input stays silent. Users who need consistent loudness across an episode should apply a dedicated loudness meter after the reverb draft is finished.
Stale results are prevented by job identity, closed AudioContext instances, and revoked object URLs, so changing the file, effect, or amount always invalidates older work in the tab. All file reading, decoding, sample processing, WAV encoding, and download creation happen in the current browser tab; audio is not uploaded to Lizely and there is no server render. Keeping an untouched original on disk lets the user restart from a clean source at any point in the chain.