Volume booster.

The slider goes to +24 dB, and Match a loudness will add up to +30 dB to reach its target. How much you actually gain depends on the file. A quiet voice memo or a lecture recorded from the back of the room has room to spare and can come up 10 or 15 dB cleanly. A finished song is usually mastered close to full scale already, so past +3 to +6 dB the limiter is doing the work and the loudest hits get flatter. As a rule of thumb, +10 dB sounds about twice as loud. The Now and After lines show the measured loudness of your file and of the result, so you can see what each step buys. It goes the other way too: drag the gain down to −12 dB, or pick a target below the file's loudness, and the file name ends in -quieter instead of -louder.

LUFS is loudness as the ear hears it, averaged over the whole file (ITU-R BS.1770, the same measure streaming services and broadcasters use); peak level only tells you the single loudest sample. Apple Podcasts asks for about −16 LUFS, within 1 dB, with peaks under −1 dB, so pick Podcast for a show. Spotify plays music at −14 LUFS by default, turning louder tracks down and quieter ones up, with −11 and −19 as listener settings, so −14 is a sensible target for music. YouTube does not publish a number. Loud (−11) and Very loud (−9) are for a file that has to stand up next to commercial masters outside a streaming app, at a cost in punch.

Yes. Drop the file and the Now line shows its integrated loudness and its peak within a moment, before you change anything; you do not have to download a thing. The number is integrated loudness as ITU-R BS.1770-4 defines it, the one Spotify, Apple and broadcasters quote: K-weighted, in 400 ms blocks, gated at −70 LUFS and again 10 LU below the average, so silence and quiet gaps do not drag it down. Two things it does not show: the peak is the highest sample, not true peak, which can read a few tenths of a decibel higher; and there is no short-term or momentary meter or loudness range, because it reads the whole file at once rather than playing it through a meter.

Not by clipping. A limiter follows the gain: it looks 5 ms ahead and turns the sound down just enough, just in time, to keep every sample under −1 dB, then lets it back up over 150 ms, so no peak is cut flat. What a big boost does change is the dynamics: the more the limiter works, the less the loud hits stand out from the rest. The note under the numbers says when the limiter is working. One caveat for delivery specs: the limiter holds sample peaks, and a true-peak meter or an MP3 decoder can read a few tenths of a decibel higher between samples.

It should not be, for files up to fifteen minutes: the tool measures the result, corrects the gain and measures again, so the After figure lands within about 0.1 LU of the target. The limiter is the reason it has to: shaving the peaks takes a little loudness off, more the harder it works. The format matters too: a WAV or a 320 kbps MP3 measures what the screen says, while a 192 kbps MP3 reads about a quarter of a decibel lower, because the encoder drops the top octave, which loudness meters weight up (we measured the same drop re-encoding an untouched file). For longer files the After figure is the arithmetic, marked "about", to keep the page responsive, and the real result can come out a few tenths quieter.

No. It makes the audio file louder, which helps when one recording is much quieter than everything else you play, but your device's volume stays capped where it was. A system-wide volume booster is an app or a setting on the device itself, not a web page.

Anything your browser can play: MP3, WAV, M4A and AAC, FLAC, OGG Vorbis and Opus in current Chrome, Edge and Firefox. The download is an MP3 at 128, 192 or 320 kbps, or a 16-bit, 44.1 kHz WAV, the same length as the original. We set no length limit; the whole file is held uncompressed while you work, about 20 MB a minute in stereo, so an hour is fine on a laptop. In Chrome on a laptop a two-minute song was measured in a twentieth of a second, matched to −14 LUFS in half a second, and encoded as a 192 kbps MP3 in under three seconds.

No. The file is decoded, measured, turned up and encoded on your device, and the download is made there. The page only counts that a file was saved, with the format and the setting, never a name or the audio.