Volume Booster

Upload it here and let the tool measure the file first. If there is headroom left, a straight dB boost raises the level with nothing else changing. If there is not, use Match Loudness — it raises the average level and lets the look-ahead limiter hold the peaks at your ceiling instead of letting them clip. The distortion people hear from other boosters comes from gain applied past the available headroom with nothing catching the peaks.

Start from the headroom figure the tool shows you: that is exactly how much you can add before the loudest moment reaches full scale. If it says 6 dB of headroom, then +6 dB is free and entirely clean. Beyond that you are into limiting, which trades dynamics for level — a few dB past headroom is usually fine, and 10 dB or more past it will noticeably flatten the track. For quiet voice recordings, boosts of 12 to 20 dB are often justified because the source is far below full scale to begin with.

Boosting applies a gain figure you choose. Normalizing measures the file and calculates the gain for you so a particular measurement lands on a target. Peak normalisation targets the loudest sample, which guarantees no clipping but says nothing about how loud the track feels. Loudness normalisation targets integrated loudness in LUFS, which tracks perception much more closely. If your goal is "make this sound as loud as my other tracks", you want loudness normalisation.

dB on its own describes signal level — how big the numbers in the file are. LUFS (Loudness Units relative to Full Scale) describes perceived loudness, measured with a frequency weighting that approximates how the ear responds and a gate that ignores near-silence. The standard behind it is ITU-R BS.1770, and LUFS is the same thing as LKFS. Two files can have identical peak levels in dBFS and differ by 10 LUFS in how loud they actually sound, which is precisely why peak measurements are a poor guide to loudness.

Because you almost certainly used peak normalisation. It finds the single loudest sample and scales the whole file until that sample hits the ceiling — so one isolated drum hit or door slam consumes the entire budget and everything else stays where it was. The EBU noted exactly this problem when it moved broadcasting to loudness normalisation: peak normalising leaves large loudness differences between programmes. Switch to Match Loudness, which measures the whole file rather than its single worst moment.

Generally no, and this is worth knowing before you crush a mix. Spotify normalises playback to around −14 LUFS: masters louder than that get turned down, and quieter ones get turned up. So a track you boosted to be louder than the target is simply attenuated on playback — you keep the squashed dynamics and lose the loudness you traded them for. Boosting is genuinely useful for local files, offline listening, phone speakers, video edits, voice memos, and DJ prep. It is not a way to be louder on a platform that normalises.

Your samples are measurements of a continuous waveform, not the waveform itself. When a converter or a lossy decoder reconstructs the curve between them, it can rise above the highest sample — so a file that measures exactly 0 dBFS can still clip on playback with no clipped sample anywhere in it. That reconstructed maximum is the true peak, measured in dBTP. EBU R 128 specifies that true peak should not exceed −1 dBTP in production and notes that lossy distribution may require lower still, which is why this tool offers −2 dBTP as well and enforces the ceiling on the reconstructed waveform rather than the raw samples.

Clipping is what happens when a sample would need to exceed the largest value the format can represent, so it is truncated flat. Those flattened tops add harmonics that were never in the recording, and the ear hears that as a crackle, buzz, or gritty edge riding on the loudest moments — usually worst on bass notes and vocal peaks. The important part is that it is permanent: turning the volume down afterwards does not undo it, because the distortion is baked into the samples.

Yes, within your headroom. Gain applied below the ceiling is a straight multiplication — nothing is discarded and no distortion is introduced. Two things can cost you quality. Boosting past your headroom engages the limiter, which compresses transients; that is a musical trade, not a defect, but it is a real change. And exporting to MP3 re-encodes the audio, which is lossy regardless of level. Export WAV if you want the boost to be the only thing that happened to the file.

Yes, and it is a hard one. Digital audio has a maximum representable value, so no file can exceed 0 dBFS — full scale is the ceiling for everyone. Once your peaks are already there, the only way to raise average loudness is to reduce the distance between the loud and quiet parts, which is what limiting and compression do. That works up to a point and then starts audibly flattening the music. Anyone promising unlimited loudness is either clipping your file or compressing the life out of it.

It watches slightly ahead of the audio you are hearing, so when a peak is coming it has already reduced the gain by the time the peak arrives, then lets the gain recover smoothly afterwards. That is the difference between a limiter and a plain compressor: a compressor reacts after the fact and lets the first few milliseconds of a transient through. Web Audio's built-in compressor caps its ratio at 20:1, so it cannot hold a hard ceiling — this tool computes the gain envelope over the samples directly instead, which is why the ceiling is exact.

Use Match Loudness with the −14 LUFS target. Voice memos and interview recordings are usually far below full scale, so there is a lot of clean headroom and the boost costs you nothing. If the recording has occasional loud moments — a cough, a bump on the mic — the limiter catches them so they do not dictate the level of the whole file, which is exactly the failure case that makes peak normalising useless for speech.

No. Gain multiplies the amplitude of each sample and leaves the timing and frequency content untouched, so the track plays at exactly the same speed and key. If you want to change those, they are separate operations — the speed changer alters tempo, and the pitch shifter transposes the key without moving the timeline.

Anything your browser can decode, which in practice covers MP3, WAV, FLAC, OGG, and M4A. Export is WAV or MP3 at 320 kbps. There is no imposed file size or length limit because nothing is uploaded — the only ceiling is what your device can hold in memory — and there is no watermark on the output.