Fading in at the start of a track and out at the end is covered in our guide to fade in and fade out. A crossfade is different: it happens between two clips, both are playing during the overlap, and what matters is how their two gain curves add up. That sum is what your ears hear, and it is where most crossfades go wrong.
What happens during a crossfade
For the length of the crossfade, the outgoing clip is multiplied by a gain that falls from 1 to 0 while the incoming clip is multiplied by a gain that rises from 0 to 1, and the two are added together. The clips overlap, so the joined track is shorter than the two clips laid end to end — by exactly the crossfade length. Two 60-second songs with a 4-second crossfade make 1:56 of audio.
The question is which pair of curves to use. There are two standard answers.
Linear vs equal-power
A linear crossfade uses straight lines: the outgoing gain is 1 − t and the incoming gain is t, where t runs from 0 to 1 through the overlap. Halfway through, each clip is at 50% amplitude. If the two clips are the same sound — two takes of one note, perfectly in phase — their amplitudes add to 100% and the level stays constant.
But two different songs are not in phase. When unrelated signals are mixed, it is their power (amplitude squared) that adds. At the midpoint of a linear crossfade the power is 0.5² + 0.5² = 0.5 — half the power, a dip of 3 dB. You hear it as the music briefly getting thinner and quieter right at the join.
An equal-power crossfade fixes this by using a quarter of a cosine for the outgoing clip and a quarter of a sine for the incoming one. Because cos² + sin² = 1 at every point, the combined power stays exactly constant through the overlap. At the midpoint each clip sits at 70.7% amplitude (−3 dB), not 50% (−6 dB).
| Position in the crossfade | Linear out / in | Equal-power out / in | Linear power sum |
|---|---|---|---|
| 0% | 1.00 / 0.00 | 1.00 / 0.00 | 0 dB |
| 25% | 0.75 / 0.25 | 0.92 / 0.38 | −2.0 dB |
| 50% | 0.50 / 0.50 | 0.71 / 0.71 | −3.0 dB |
| 75% | 0.25 / 0.75 | 0.38 / 0.92 | −2.0 dB |
| 100% | 0.00 / 1.00 | 0.00 / 1.00 | 0 dB |
The equal-power power sum is 0 dB on every row, which is the point. Use equal-power for nearly everything: song into song, music into speech, ambience into ambience. The only case where linear is better is a crossfade between two copies of the same recording (repairing an edit inside one take), where equal-power would produce a 3 dB bump instead. The audio joiner uses equal-power crossfades for that reason.
How long should a crossfade be?
| Situation | Length | Why |
|---|---|---|
| Joining takes of speech | 0.05–0.3 s | Just enough to remove the click; nobody hears an overlap |
| Sound effects in sequence | 0.1–0.5 s | Keeps transients sharp |
| Music bed into voice | 1–3 s | Music recedes as the voice arrives |
| Song into song (playlist, workout mix) | 3–6 s | Feels continuous without muddling two choruses |
| Ambient or slideshow music | 4–8 s | Slow material tolerates long overlaps |
Two practical limits. First, a crossfade can’t be longer than the clips it joins: if a jingle lasts 4 seconds, it can share at most 2 seconds with each neighbour, or it would be overlapping both at once. Our joiner shortens such crossfades automatically and says so. Second, long crossfades between songs with strong beats sound messy when the tempos differ, because two drum patterns play at once. Start the crossfade where the outgoing song is already winding down — its last chorus fading, the drums stopped — by trimming the clip’s end, rather than making the crossfade longer.
Recipe: a podcast intro in five minutes
A classic intro is a short music sting that swells, settles, and gives way to the host’s voice. You need three files: an intro music track you have rights to (royalty-free libraries are fine), the episode recording, and optionally an outro.
- Trim the music. Add the music to the joiner and drag across its waveform to keep 8–15 seconds — start on a downbeat and end shortly after a phrase finishes, so the crossfade begins as the music relaxes.
- Trim the voice. Add the episode and trim off everything before the host’s first word, leaving about 0.2 seconds of room tone.
- Set the crossfade. Choose Crossfade and 2–3 seconds. The music fades under the first words instead of stopping dead.
- Balance. Lower the music clip’s gain by about 6 dB so the first words aren’t buried. Play the join and adjust until the voice is clearly on top.
- Add the outro. Add the outro music as a third clip with the same crossfade, or switch the last join to a 1-second gap if you prefer a clean stop before the music.
- Export. Leave Normalize on, tick Mix to mono for a speech show, and download MP3 at 128 kbps (mono) or 192 kbps (stereo).
Our joiner doesn’t do automatic “ducking” (lowering the music only while someone speaks), so in this recipe the music and voice overlap only during the crossfade. That is the cleanest-sounding option for a short intro anyway; music running under the whole first minute is a job for a multitrack editor. If you want the music to fade in from silence at the very start, set Fade in at start to 1–2 seconds.
Common crossfade mistakes
- Crossfading two loud choruses. The overlap doubles the density and can clip. Trim so the outgoing clip is already quieter, and keep Normalize on.
- Using a crossfade to hide a bad edit. A 5-second crossfade over a cut mid-sentence still sounds like two sentences talking over each other. Cut speech in pauses.
- Forgetting the length change. Every crossfade shortens the result. If the finished file must hit an exact duration — a 30-second ad, a slideshow — account for it, or use gaps.
- Linear curves between different songs. The 3 dB dip is subtle on headphones and obvious on a PA. Equal-power costs nothing.
For the full workflow of combining files — ordering, levels, formats — see how to merge audio files.