Every file gets its own gain, worked out from its own average level, so the loud ones come down and the quiet ones come up. A peak ceiling sits on top, so a file that cannot reach the target without distorting stops short and says so rather than clipping. Handed back as WAV.
Free · No account · Your files never upload
Levelled on your computer and handed back as WAV. Nothing is uploaded.
Works out the gain each file needs to reach a common average level, applies it, and hands everything back as WAV. Files that were stereo with two identical channels can be folded to mono on the way through.
Raising a quiet file to the average level can push its loudest moments past full scale, which distorts. So the gain is whichever is smaller: what the average target asks for, or what the peak ceiling allows. A file that stops short is marked as capped rather than quietly clipped, and that mark usually means the recording has a few very loud moments against a quiet average and wants compression before levelling.
The browser can decode almost any audio format and will only encode this one. That is the right answer anyway: WAV is lossless, every authoring tool imports it, and Storyline or Rise re-encodes on publish regardless. A lossy intermediate would stack one generation of loss on another for no benefit.
It applies one gain across each whole file. It does not compress, gate or de-noise, all of which change how the recording sounds rather than just how loud it is, and all of which belong in an audio editor where you can hear what you are doing.