Before using audio effects online, confirm that one browser-decodable file is no larger than 50 MiB, choose one disclosed effect, and understand that the export is a new PCM16 WAV rather than an edited copy of the source container.
Audio-effect names can suggest much more complex processing than a simple browser tool performs. For a bounded local workflow, Audio Effects Online applies one defined transformation at a time. File reading, decoding, sample processing, WAV encoding, and download creation happen in the current browser tab. Audio is not uploaded, stored in an account, rendered on a server, or transformed by a hidden AI process. That simplifies the privacy decision, but it does not remove browser compatibility or resource limits.

Understand the local, one-effect workflow
Treat each run as a controlled edit with a clear beginning and end:
- Input: one browser-decodable audio file no larger than 50 MiB.
- Operation: one of echo, lightweight reverb, reverse, fade in, fade out, or normalization.
- Strength: one whole-number amount from 0% through 100%.
- Result: a complete, interleaved, little-endian PCM16 WAV downloaded as a new file.
The amount changes the selected operation rather than choosing a different effect. Your source file is not overwritten, but keeping an untouched copy remains important, especially when planning several edits or comparing results.
Match the effect to its actual behavior
Choose an effect by the sound change it makes, not by a vague promise such as “studio quality” or “professional space.”
| Effect | What the amount controls | Practical use | Important boundary |
|---|---|---|---|
| Echo | Scales a single delayed copy; at 100%, it adds a half-strength copy 200 milliseconds later. | Simple repetition or a draft delay effect. | It is one reproducible echo, not a multi-tap delay or a full delay processor. |
| Lightweight reverb | Scales three disclosed early reflections. | A quick sense of space for a sound-design draft. | It does not simulate a room, studio, plate, spring, or convolution reverb. |
| Reverse | Blends the original with a version whose sample frames are in the opposite order. | Reversing a passage or creating a blended transition. | The output remains a blend at amounts below 100%, rather than becoming fully reversed. |
| Fade in | Sets the dry proportion at the beginning and ramps linearly toward full level at the end. | An opening fade or gradual entrance. | It is a full-range linear fade, not a timed automation curve. |
| Fade out | Sets the dry proportion at the end and applies the inverse of fade in. | An ending fade or gradual disappearance. | It is a full-range linear fade, not a timed automation curve. |
| Normalization | Scales the largest absolute decoded sample toward a 0.95 sample peak. | Adjusting peak level while retaining the recording's existing balance. | It is not LUFS or perceived-loudness matching. |
At 100% amount, echo retains every input sample and adds a half-strength copy 200 milliseconds later. Lower amounts scale the delayed copy linearly. Reverb adds copies at 29.7, 37.1, and 41.1 milliseconds with disclosed weights of 0.50, 0.35, and 0.25 multiplied by the selected amount. These are product design choices, not measurements of a concert hall or professional processor. For a closer distinction, see how the lightweight reverb differs from room and studio emulation.
The lightweight reverb does not model room dimensions, frequency-dependent decay, stereo diffusion, damping, pre-delay, convolution, or a professional reverberation tail. It can add space for a quick draft, but it should not be presented as a realistic acoustic simulation.
Check compatibility before processing
The browser's Web Audio implementation performs the decoding. MP3, WAV, M4A, AAC, Ogg, WebM, and FLAC filenames are accepted, but actual codec and profile support varies by the browser and operating system. A recognized filename does not guarantee that the encoded content will decode. The W3C Web Audio API 1.1 defines the decoding contract and the AudioBuffer sample-frame model used by this workflow. If decoding fails, the tool creates no result.
- The encoded source file must be no larger than 50 MiB.
- Decoded audio must contain 1–8 equal-length channels.
- The decoded sample rate must fall between 8,000 and 192,000 Hz.
- Decoded data must fit both the declared duration budget and a ceiling of 30 million channel samples.
- The complete processed output must also remain within 30 million channel samples.
File size and decoded length are separate checks. A compressed file can be below 50 MiB while still producing more than 30 million decoded channel samples. See what the 50 MiB limit means for Audio Effects Online for the distinction between encoded size and decoded workload.
Echo and reverb can pass the input ceiling but exceed the output ceiling because their tails add frames. When that happens, the complete tail is not truncated and no output file is created.
Apply one effect step by step
- Choose and protect the source. Select one browser-decodable audio file no larger than 50 MiB. Keep an untouched copy of the original elsewhere so you can compare versions or start again without stacking an unwanted result.
- Select one effect. Choose echo, lightweight reverb, reverse, fade in, fade out, or normalization. The available effects are separate choices; the tool does not run all six at once.
- Set the amount. Enter a whole-number value from 0% to 100%. The amount blends or scales the chosen operation, so record the value if you need to explain or reproduce the result.
- Apply the effect locally. Start the local operation after confirming the file and controls. For echo or reverb, the complete tail must fit the output budget. If it does not, the tool creates nothing rather than shortening the result.
- Inspect the reported values. Review the channel count, sample rate, input and output frame counts, output peak, and exact WAV byte size. Echo and reverb can produce more output frames than input frames because the delayed tail is included.
- Download the complete WAV. The result is a new interleaved, little-endian PCM16 RIFF/WAVE file. It is not the original MP3, M4A, AAC, Ogg, WebM, FLAC, or WAV container with tags and artwork carried over.
- Use the result deliberately in any chain. To add another effect, download this WAV, inspect it, and select it as the input for a later run. Changing the file, effect, or amount invalidates older work, so download and record the current result before changing controls.
Read peak, clipping, and fades correctly
The reported output peak should not be interpreted as a loudness measurement. Normalization finds the largest absolute decoded sample and scales toward a 0.95 sample peak; silent input remains silent. It does not target LUFS, perceived volume, broadcast compliance, dynamics, inter-sample true peak, or equal loudness across different recordings.
Strong echo and reverb settings can clip when dry and delayed samples overlap. Finite output samples are constrained to the legal -1 through +1 range before WAV quantization. This prevents an illegal floating-point result, but clipping can still change dynamics and flatten peaks. The final WAV also quantizes floating samples to signed 16-bit PCM.
Reverse and the two fades work through blending or scaling, so the amount changes how much of the transformation is present. Always judge the result against the source rather than assuming that the strongest setting produces the most useful recording.
Plan an effect chain without losing the source
Only one effect is applied per run. Building a chain therefore means downloading a completed WAV, inspecting it, and using that WAV as the next input. Reapplying the tool to its own PCM output is safe as a separate operation, but repeated lossy decoding and PCM processing may reveal source artifacts. Repeated clipping or normalization can also alter dynamics, so avoid stacking an intermediate result that already sounds wrong.
The export preserves the processed channel samples and sample rate, but it creates a different file type. It does not preserve the source codec, bitrate, compression quality, tags, artwork, chapters, cue points, loop markers, or other container metadata. Keep any metadata you need separately.
Microsoft's WAVEFORMATEX documentation defines PCM block alignment as channels multiplied by bits per sample divided by eight. At 16 bits per sample, that means two bytes for each mono frame and four bytes for each stereo frame. Use the tool's exact output byte size when checking the final file rather than estimating it from the source file size.
Make a careful first export
Before treating the result as finished, run through this short checklist:
- Confirm that the original remains available in an untouched form.
- Record the selected effect and amount so the choice is clear.
- Compare the input and output frame counts, especially for echo or reverb.
- Review the output peak for evidence of clipping, while remembering that peak is not perceived loudness.
- Check the exact WAV byte size and remember that the export is a new PCM16 file.
- Start a chain only from an intermediate result you have inspected and deliberately chosen.