Guide
Making audio louder: gain, peak normalising and limiting
A recording made too quietly is a common problem: a lecture from the back of a hall, a voice memo with the phone in a pocket, a track exported from a program with a low master fader. The fix is gain, and the question is how much. Too little and nothing has changed; too much and the loudest moments break up. This booster calculates the safe amount for you, shows what will happen, and writes the result in the format of your choice.
Gain and decibels
A gain of +6 dB doubles every sample value, +12 dB quadruples it, and −6 dB halves it. The scale is logarithmic because hearing is: equal steps in decibels sound like equal steps in loudness. A digital file has a hard ceiling at full scale. Any sample that the gain pushes beyond it cannot be stored, so it is clipped flat, which sounds like crackling distortion.
Normalising the peak
Peak normalising looks for the highest sample in the file, on any channel, and chooses the gain that brings that peak to a target. The target defaults to −1 dBFS, which leaves a little headroom for the rounding that lossy encoders do. Because only the single highest peak is considered, one stray click can stop the whole file from getting louder; if the result is quieter than you hoped, look for the spike on the waveform. Normalising changes level, not perceived loudness: two files with identical peaks can still sound very different in volume.
A worked example
Press Sample for a stereo arpeggio. The page reports a highest peak of −3.65 dBFS, a gain to apply of 2.65 dB and a highest peak after of −1 dBFS. Apply writes a WAV whose largest sample is 0.8912, which is 10 raised to −1/20, exactly the target. Switch to Add gain and set +6 dB: the page now predicts a peak above 0 dBFS (−3.65 + 6 = +2.35) and warns that the loudest parts will be rounded off or clipped.
The limiter
When the gain would push samples over full scale, Round off the peaks applies a smooth curve: values up to 0.8 are untouched, and above that they bend gradually towards 1.0 without reaching beyond it. Hard clip simply cuts at ±1. Soft limiting usually sounds gentler but still distorts, and the note under the result shows how many samples were limited. If nothing would exceed full scale, neither mode touches the sound, so a plain gain stays an exact gain.
Edge cases
Silent files cannot be normalised (there is no peak), and the tool leaves them unchanged. Boosting a quiet recording also boosts its noise and hiss, so the improvement is limited by how clean the original is. Stereo files get the same gain on both channels, which keeps the balance. Formats that compress, Opus and AAC, can add small overshoots after encoding, which is why a peak just below zero is safer than exactly zero.
Limits
The tool does not compress dynamics, apply an equaliser or measure loudness; use the LUFS meter to see how loud a file really seems. It cannot recover clipped audio. It works on whole files and runs on your device.
Related tools
- LUFS Loudness Meter measures perceived loudness before and after.
- Audio Cutter removes a loud spike before normalising.
- Mic Test shows whether a quiet recording is a gain problem at the microphone itself.
- Audio Converter changes the format of the boosted file.
Checked against the current version of the Audio Volume Booster & Normalizer on .