Audio Volume Booster

A recording made too far from the microphone, a voice memo taken across a room, an interview captured at a cautious level — all of them are perfectly good audio sitting too low to hear comfortably. Boosting multiplies every sample by the same amount, which raises the whole recording without changing anything about how it sounds, right up until the loudest moments run out of room.

• Voice and podcasts: bring quiet speech up to a level people can listen to without reaching for their volume control.
• Music: lift an under-recorded track before you mix it or send it on.
• Video sound: rescue dialogue that was recorded too low to sit under music.
• Reducing, too: a negative gain pulls a recording that is already too hot back down.

Headroom is the distance between the loudest moment in your recording and the level you are willing to let it reach. If a file peaks at -20 dBTP and your ceiling is -1 dBTP, it has 19 dB of headroom: boost by 19 and it just touches the ceiling, boost by 20 and it goes over. Every question about how much to boost is really this question, and it has an exact answer for every file.

• The file is measured the moment it lands, before you choose anything — its true peak, its loudness, its sample peak and its RMS.
• Maximum without clipping sets the gain to the headroom, so the largest safe boost is one press away.
• The ceiling is a TRUE peak, sampled four times per sample. A lossy encoder moves peaks between samples, and a ceiling that cannot see them is not a ceiling.
• Ask for more than the headroom and the guard holds the ceiling, then reports exactly how much gain reduction that cost.

Applying a gain is easy to claim and easy to get wrong. So after the file is written, it is decoded and measured a second time with the same meter that measured the source, and the difference between the two is the gain that was actually applied. That number appears beside the one you asked for.

• Loudness is scale-invariant, so a correctly applied gain of +6 dB moves a file's loudness by exactly 6 dB — for any material, with no dependence on content.
• That makes the check exact rather than approximate: a run that lands more than half a decibel away fails instead of handing you a file that quietly did not get the boost.
• The verified true peak is shown too, so you can see the ceiling was held rather than take it on trust.
• No competing online booster measures your file before, and none measures the file it wrote afterwards.

Clipping is what happens when a sample is asked to go past full scale and has nowhere to go: the waveform's peaks are flattened off, which is heard as a harsh, brittle edge on the loudest parts. It is not reversible, so the only useful moment to deal with it is before the file is written.

• Hold the ceiling: a true-peak limiter engages only when the gain would actually breach the ceiling, and only then. A limiter that runs on every boost is indistinguishable from one that does nothing.
• Let it clip: sometimes that is genuinely what you want — a voice memo bound for a phone speaker, where audibility beats fidelity. It is never silent: the tool warns you and reports the peak it landed on.
• Lower the ceiling: -1 dBTP is the streaming convention and leaves room for a lossy encoder to move peaks. -2 is the cautious choice if the file is going through another encode after this one.

They answer different questions and the choice is usually obvious once the question is stated. A booster is told the gain and works out what will happen. A normalizer is told the outcome and works out the gain.

• Use the booster when you know how much louder you want this file, or when you want the largest lift it can take.
• Use the normalizer when several files need to end up at the same loudness as each other, or when a platform publishes a target you have to hit.
• Use both when a file needs a specific lift now and a delivery target later — boost first, normalize last.

• Start with Maximum without clipping. It is the largest boost the file supports, and everything else is a decision to take less than that.
• Reduce noise first, not after. Gain multiplies hiss along with the voice, so removing it before the boost keeps it out of the result.
• Export lossless for anything you will edit again. WAV and FLAC keep the boost exact; a lossy format spends a generation of quality on it.
• Watch the applied figure, not the request. If the guard engaged, the applied gain is smaller than the one you asked for, and that number is the truth about your file.

Three passes over your file, all of them on your device. Nothing is uploaded at any point, and nothing needs to be — the whole pipeline runs where the file already is.

• Measure. The file is decoded once and its true peak, loudness, sample peak and RMS are read off it. That is where the headroom figure comes from.
• Apply. The gain is applied, the clip guard is inserted only if the arithmetic says it is needed, and the file is encoded in the format you chose.
• Verify. The finished file is decoded and measured again, and the applied gain is reported against the requested one.

Powered by optimised audio processing algorithms.

• Audio Normalizer
• LUFS Loudness Meter
• Noise Reduction
• Audio Cutter
• Audio Merger
• Audio Converter