A hard cut at the start of a clip produces a click, because the waveform jumps from silence to a nonzero value in a single sample. Your ears hear that discontinuity as a tick, and cheap speakers exaggerate it. A fade of even fifty milliseconds removes the click completely, which is why every podcast, every advert and every music release has one whether or not you notice it.
Everything runs locally through the Web Audio API. The file is decoded to raw samples, each sample near the head and tail is multiplied by its gain value, and the result is written out as a 16 bit PCM WAV. There is no upload, no queue and no account.
When to add a fade to an audio file
Short fades are hygiene. Long fades are a creative decision. The two lengths solve completely different problems and mixing them up is the most common mistake people make with fades.
Anything under about a fifth of a second is inaudible as a fade and simply removes the click at a boundary. Anything over a second reads as a deliberate musical gesture, so it belongs at the end of a song or under a voice over, not on a sound effect.
- 20 to 50 milliseconds to kill a click at a cut point without altering the sound.
- 200 to 500 milliseconds to soften the entry of a sound effect or a stinger.
- 1 to 3 seconds to close a podcast segment or duck out of a music bed.
- 5 to 10 seconds for the classic long fade out at the end of a track.
- Match the fade in and fade out lengths when a clip will loop, so the join is symmetrical.
Choosing the fade curve
A fade is a gain value that moves from zero to one over time, and the path it takes between those two numbers is the curve. Linear moves in a straight line, which sounds fine over short lengths but noticeably fast in the middle over long ones, because loudness perception is roughly logarithmic rather than linear.
The equal power curve follows a quarter of a sine wave. It holds more level through the middle of the fade, which is why it is the standard choice for crossfades and for long musical fade outs. The exponential curve does the opposite, staying quiet longer and then rising sharply, which suits a sound that should creep in from nowhere.
If you are unsure, use equal power for anything longer than a second and linear for anything shorter. Below a quarter of a second the difference between the three curves is inaudible.
How this audio fade in out tool applies the envelope
The gain is applied to the sample values, not to a playback node. Once decoding is done the tool walks each channel array, computes a gain factor for the position, multiplies the sample by it and writes the product into a new buffer. The fade is therefore permanent and travels with the file into any editor or player.
That matters because the alternative approach, automating volume in a session, only exists inside the project that created it. Exporting a stem with the fade already printed means the person who receives it cannot accidentally lose it.
Both fades are computed independently and combined by taking the lower of the two gains at any position. If you set lengths that overlap, the file simply gets quieter through the middle rather than producing a discontinuity, and the tool warns you before that happens.
Why the download is a WAV file
Browsers include decoders for MP3, AAC, OGG and FLAC but they include no encoder for those formats, so a compressed download would require shipping a codec into the page first. WAV can be written directly by laying out a RIFF header and appending 16 bit PCM samples.
For fade work that is the right output anyway. Fading multiplies sample values, and re encoding a lossy file after every edit compounds generation loss. Print the fades to WAV, do any further editing, and compress once at the very end.
Fades, loops and gapless playback
Fades and seamless loops pull in opposite directions. A loop needs the last sample to lead naturally into the first, and a fade out followed by a fade in creates an audible dip at the seam. For looping material, use fades of a few milliseconds only, or none at all if the source was recorded at a zero crossing.
Album style gapless playback has the same constraint. If two tracks are meant to run into each other, fading the first out and the second in inserts a hole that no player can remove afterwards. Keep smooth audio fades for material that stands alone.
| Length | Perceived as | Typical use |
|---|---|---|
| 20 to 50 ms | Inaudible | Removing a click at a cut point |
| 200 to 500 ms | A soft entry | Sound effects and stingers |
| 1 to 3 s | A deliberate gesture | Podcast segment endings |
| 5 to 10 s | A musical fade | Song outros and music beds |
How to fade audio with this audio fade in out tool
- 1
Load your audio
Drop an MP3, WAV, OGG, FLAC, M4A or AAC file onto the panel. It is decoded in the page and never uploaded anywhere.
- 2
Set the two fade lengths
Enter the fade in and fade out durations in seconds. The tool caps them at the file length and warns if they overlap.
- 3
Pick a curve
Linear for short fades, equal power for long musical ones, exponential for a sound that should creep in slowly.
- 4
Render and download
Press apply, listen to the preview player, then download the WAV with the envelope written into the samples.
Related tools worth bookmarking
Sources and further reading
Frequently asked questions
Common questions about the audio fade in out tool.
No. Only the samples inside the two fade regions are multiplied by a gain factor. Everything between the end of the fade in and the start of the fade out is copied across untouched, so the body of the track keeps its original level exactly.
A cut that lands on a nonzero sample makes the waveform jump instantly from silence to that value. That instantaneous step contains energy across the whole frequency range, which your ears hear as a click. A fade of twenty milliseconds smooths the step out of existence.
Yes, but if the two lengths add up to more than the file duration they will overlap and the middle of the clip will be quieter than the source. The tool detects that case and shows a warning so you can shorten one of the two values.
No. Decoding, envelope application and WAV writing all run inside your browser tab using the Web Audio API. Nothing is transmitted, nothing is stored on a server and no one else can see the file. Watch the network panel during a render to confirm it.
Equal power, over five to ten seconds. Linear fades of that length sound like they rush through the middle because human loudness perception is closer to logarithmic than linear, so the level appears to drop faster than the number does.