Audio Fade In / Fade Out Tool

Audio Fades Explained: How to Apply Professional Fade In and Fade Out to Your Recordings

A discontinuity at the start or end of a digital audio file produces an audible click that can reach -3 dBFS on a peak meter. This happens because the playback system jumps instantaneously from silence (zero amplitude) to the first sample value of the audio, or from the last sample value back to silence. The steeper that jump, the louder the click. Fades solve this problem by ramping the amplitude gradually from zero to full volume (fade in) or from full volume to zero (fade out) over a defined duration. Every professional DAW applies fades automatically when you trim or bounce audio. This tool lets you apply them to any file without installing software.

What Are Audio Fades and Why They Matter

A fade is a volume envelope applied to the beginning or end of an audio region. The fade multiplies each sample by a gain value that changes over time, starting at zero and reaching 1.0 (full volume) for a fade in, or starting at 1.0 and reaching zero for a fade out. Without a fade, the abrupt transition between silence and audio (or audio and silence) produces a broadband click. The click is caused by the sudden step in the waveform, which the DAC reproduces as a short transient containing energy across the entire audible spectrum.

Fades also serve musical purposes. A long fade out at the end of a song creates a natural sense of closure. A short fade in at the start of a recording removes the count-in noise (footsteps, breathing, chair creaks) without cutting the first note. A 50 ms fade on a chopped sample prevents clicks at each slice point while preserving the attack transient. These are all standard practices in professional mixing and editing.

Fade Curves: Linear, Logarithmic, and Exponential

The shape of the fade curve determines how the volume changes over time. This tool offers three curves.

Linear changes volume at a constant rate. The gain increases (or decreases) by the same amount per millisecond throughout the fade. This produces a perceptually uneven fade because human hearing is logarithmic: a linear fade sounds like it starts fast and ends slow. Linear fades are best for short durations (under 100 ms) where the perceptual unevenness is not noticeable, or for sample chopping where you need a predictable, mathematically simple ramp.

Logarithmic changes volume faster at the start and slower at the end. The gain follows a log curve, which closely matches how human hearing perceives volume changes. This produces a fade that sounds smooth and even from start to finish. Most DAWs use logarithmic (or similar) curves as their default fade shape. Use logarithmic for fade outs at the end of songs, where you want the volume to appear to decrease steadily to the ear.

Exponential changes volume slowly at the start and rapidly at the end. The gain follows a squared curve (progress raised to the power of 2). This creates a fade that seems to hang at full volume for most of the duration, then drops away quickly at the very end. Exponential fades are useful for dramatic endings or for fade ins where you want the audio to seem to appear suddenly from silence.

How This Tool Applies Fades

When you load a file, the tool decodes it using the Web Audio API's decodeAudioData() function, which produces an AudioBuffer containing raw Float32Array sample data for each channel. The fade is applied by iterating over every sample in the buffer and multiplying it by a gain value computed from the fade curve function. For a fade in, samples from index 0 to fadeInSamples are multiplied by getFadeGain(i / fadeInSamples, curve). For a fade out, samples from fadeOutStart to the end are multiplied by getFadeGain(1 - progress, curve). If both fades overlap (the file is shorter than fade in plus fade out), the gains are multiplied together, producing a smooth crossfade through the middle.

The processed buffer is cached and only recomputed when you change a fade parameter. The A/B preview lets you switch between the original and processed buffer instantly, so you can hear exactly what the fades do before you commit to the export. The export encodes the processed buffer as a 16-bit PCM WAV file at the original sample rate.

Recommended Fade Durations for Common Scenarios

For song endings, 2 to 5 seconds is standard. A 3-second fade out gives the listener time to process the ending. Shorter than 2 seconds can feel abrupt. Longer than 8 seconds is unusual unless the song is designed for it (ambient, drone, or classical recordings).

For voiceover and podcast editing, 0.5 to 1 second is typical for the start of a clip. This removes the abrupt entry of the voice without being long enough to sound like a deliberate effect. For the end of a voice clip, 0.3 to 0.5 seconds smooths the transition to silence.

For sample chopping and loop creation, 5 to 50 ms is the right range. The goal is to remove the click at the slice point without softening the attack transient. Use the Audio Looper after fading your samples to set seamless loop points with zero-crossing snap.

For removing count-in noise (footsteps, breathing, chair sounds before the first note), 100 to 300 ms is sufficient. The fade brings the volume from zero to full before the first note hits, masking the ambient noise. If you need to remove longer silences entirely, use the Silence Trimmer first, then apply a short fade to clean up the resulting cut points.

Using Fades with Other Tools

After applying fades, you may want to check the loudness of the result. The LUFS Loudness Meter measures integrated loudness and true peak, which tells you if your faded file meets streaming platform targets. If you need to trim the file before fading, use the Audio Trimmer to isolate the section you want. For pitch-shifting a faded file, the Pitch Shifter handles transposition without affecting the fade envelopes. And if you recorded the audio yourself, the Audio Recorder captures microphone input directly in the browser.

The AES (Audio Engineering Society) recommends fades on all digital audio edits in their AES60 standard for digital audio production. The principle is simple: any discontinuity in the waveform produces a click, and a fade is the most computationally efficient way to remove that discontinuity. For technical detail on how gain envelopes interact with sample data, the Web Audio API specification on GainNode documents the underlying gain multiplication algorithm, and the AES60 standard covers best practices for fade application in digital audio production workflows.