Add voice notes, song parts or podcast segments, put them in order and choose how each one meets the next. Quiet clips are levelled up to match the loud ones. Everything is mixed in your browser and saved as a WAV.
The volume curves show what happens across a two-second join. The right choice depends on what you're joining: talk, music or recordings from different places.
Joining raw files byte by byte only works when every clip has the same format. Usually they don't, so each one is decoded, adjusted to a shared setting, then mixed. These are the differences we reconcile for the clips above.
A straight cut joins whatever sample one clip ends on to whatever the next begins with. If those values differ, the speaker cone has to jump instantly, a step that contains every frequency at once, heard as a click. Below, a 220 Hz tone is cut into a 330 Hz tone three ways, and the energy the join adds above 4 kHz is measured.
When two unrelated sounds overlap, their powers add. When two identical sounds overlap, their waves add. A linear fade keeps the waves constant but loses power in the middle; a sine/cosine “equal-power” fade keeps power constant but overshoots when both clips are the same. Measured on 1-second fades:
Phone recordings are often 48 kHz and music 44.1 kHz, so joining them means resampling one. A 48 kHz test tone sweeping from 0 to 24 kHz is converted three ways. Anything above 22.05 kHz can’t exist at 44.1 kHz; a careless converter reflects it back as a falling tone that was never in the source.
Lossy encoders add silent “priming” samples at the start and pad the end to a whole frame. Joining decoded MP3 or AAC files without trimming those leaves tiny gaps, the reason gapless albums need metadata. Values are typical for common encoders, from general knowledge of their documentation.
The joined file is lossless; these tools shrink or finish it.