Fade In & Out

Ramp a recording up from silence at the start and back down at the end, without the file leaving your machine.

Runs on your deviceAudioWorkingNo account needed

What this tool does

Fade In & Out multiplies the samples at each end of a recording by a rising and a falling ramp, so the track lifts out of digital silence and settles back into it. Each ramp begins on zero exactly, and everything between the two ramps is left alone — the samples in the middle are handed to the encoder exactly as they were decoded. All of it happens in this browser tab: WAV is read by the page's own RIFF reader, every other format is handed to the browser's decoder, and the file is not sent anywhere.

Three curves are offered, and they differ where it is audible: in the middle of the ramp. Linear raises the level evenly, which is the predictable choice for speech and for very short fades. Smooth follows a quarter sine, so it is already past half volume at the midpoint and holds on a little longer before letting go, which is why it is the default for music. Exponential squares the ramp — slow to start, arriving late — and suits a long lift under a voiceover where the bed should stay out of the way.

The ramps are anchored to the ends of the file and nowhere else. There is no fade at a timestamp in the middle, and nothing here lengthens or shortens the recording. Set both fields to zero and the run is refused rather than handing back a copy of what you gave it. Ask for more fade than the file can hold — twenty seconds at each end of a thirty second clip — and the two ramps share the length in proportion instead of overrunning, so that clip is given fifteen seconds at each end and its duration is unchanged. Worth knowing in that case: the summary line still prints the lengths you typed, and it is the note appended to it, shortened to fit, that tells you the real ramps were smaller.

A few things about the output are worth knowing before you download. The format list stops at WAV and MP3, because browsers ship decoders for AAC, FLAC and OGG but no encoders for them, so a FLAC or an M4A is read happily and written back as WAV; only an MP3 stays an MP3 under “Match the input where possible”. That MP3 route re-encodes, which costs one lossy generation on top of the original, and the file is written from the audio alone, so ID3 tags and cover art do not survive the trip. Queue several tracks and each is decoded and faded in turn, one download per input — but every file in the queue takes the same two ramp lengths and the same curve, and the details panel afterwards describes only the last one.

How to use it

  1. Add the tracks you want faded. MP3, WAV, M4A, AAC, FLAC, OGG and Opus are all accepted, as far as your browser can decode them. A queue is worked through one file at a time.
  2. Set the fade in and the fade out. Each runs from 0 to 120 seconds, in tenths. Two seconds apiece is the starting point; leave one at zero to treat only the other end.
  3. Pick a curve. Linear for speech and de-clicking, Smooth for music, Exponential when the level should stay low for longer before it rises.
  4. Choose the output format. Match the input where possible keeps an MP3 as MP3 and writes anything else as WAV. Forcing WAV offers 16-bit, 24-bit or 32-bit float; forcing MP3 offers seven bitrates from 64 up to 320 kbps.
  5. Download the result. Every file comes back named after its input with -faded added. The summary quotes the fade lengths you typed, so read the shortened-to-fit note when the two ramps did not both fit.

Processing & privacy

Runs on your device. Your file is processed directly in this browser.

Common use cases

A song that ends on a hard cut

Three seconds of smooth fade at the tail turns an abrupt stop into an ending, which matters most when the track runs straight into the next one.

Trimmed clips that click

Cutting mid-waveform leaves a step that a speaker reproduces as a click. A tenth of a second of linear ramp at each end removes it without audibly shortening the clip.

A podcast intro and outro

Bring the theme up under the opening line and down again at the close, so the music arrives and leaves rather than being switched on and off.

Background beds under a voiceover

A long exponential fade-in keeps the bed quiet while somebody is speaking and lets it climb into the gap once they stop.

Loops for exhibitions or hold music

Fading both ends to silence hides the seam where the file restarts, because the join becomes silence meeting silence.

Questions

Is my audio file uploaded?
No. Decoding, the fade and the re-encoding all happen inside this browser tab, and nothing about the audio is transmitted or stored elsewhere. The only thing still fetched from this site is the MP3 encoder itself, pulled in the first time a run asks for MP3 output.
Can I fade somewhere in the middle of a track?
No. The two ramps are fixed to the first and last seconds of the file. To soften an internal edit, cut the section with Trim Audio and fade the piece you keep, or rejoin pieces with Merge Audio, which offers a crossfade at each seam.
What happens if the fades are longer than the file?
They share it in proportion rather than overlapping. A twenty second fade in and out on a thirty second clip becomes fifteen seconds at each end, and the length of the file does not change. The summary still shows the twenty you asked for, followed by a note that it was shortened to fit.
Which curve should I choose?
Linear for speech and for short de-click fades, Smooth for music, Exponential when a slow start is the point. Halfway through a fade-in, Smooth sits louder than Linear and Exponential sits quieter — that difference is the whole of the choice.
Does fading change the length or the level of the rest?
Neither. Only the samples inside the two ramps are scaled, everything between them is copied untouched, and no time is added or removed at either end.
What format does the faded file come back as?
WAV or MP3. Match the input where possible follows the file you gave it, so an MP3 returns as MP3 while a FLAC, M4A or OGG returns as WAV — a browser can read those formats but cannot write them.
Do the tags survive?
No. The output is written from the audio alone, so title, artist, album and artwork all come back blank. Audio Metadata can put the text tags back afterwards, but only onto an MP3, and it does not restore cover art.
Can I fade several files at once?
Yes, with one caveat: the settings are shared. Each queued file is faded in turn and arrives as its own download, but they all get the same two ramp lengths and the same curve, so a half-second de-click and a five-second music fade have to be run as two separate batches.

Related tools