Make an audio file louder or quieter by an exact amount, with the clipping counted first
Processed in your browser. Nothing is uploaded. Updated on 24 Oct 2026.Changing volume here multiplies every sample by one figure, measured and applied inside the page. Your file's peak, average level and remaining headroom are shown first, and any gain past that headroom reports the exact number of samples it would flatten. Plus 6 dB doubles amplitude; plus 10 sounds twice as loud.
Press Choose an audio file or drop it onto the page; it is measured on your device and not uploaded.
Read Loudest sample, Average level and Headroom before clipping for the file.
Under Change pick By decibels and drag the slider, or Set the peak and choose a Peak at level.
Check any Before you press it warning, press Make it louder or Make it quieter, then Download the WAV.
Add a file and the page measures it before offering to change anything: the loudest single sample, the average level, and how many decibels fit above the loudest moment. Then choose either a change in decibels or a target for the peak, and the whole file is multiplied by that one figure.
A digital audio file has a ceiling. Every sample is a number between minus one and one, and there is nothing above one. Raising a file whose loudest moment already sits at minus three decibels by six decibels pushes its peaks three decibels past the ceiling, and those samples are flattened into a straight line, which is heard as distortion and cannot be repaired afterwards.
Most volume boosters do exactly that and say nothing. Here the number of samples that would be flattened is counted, exactly, before anything runs. A few dozen out of five million is inaudible and the page says so. Ten per cent of the file is ruined audio and the page says that too, along with the largest change this particular file takes cleanly.
The peak decides what clips. The average, measured as RMS, is much closer to what people mean by loud: a heavily compressed pop track and a quiet spoken recording can share the same peak and be twenty decibels apart in the way they sound. Both figures are shown, in dBFS, where zero is full scale and every real signal is negative.
Loudness matching in the broadcast sense, measured in LUFS, is a different job that needs compression to do properly, and compression changes a performance. Nothing here compresses anything.
One WAV, at 16 or 24 bits, because no browser can write an MP3. Nothing is re-compressed: each sample is decoded, multiplied and written. The file is therefore much larger than a compressed original, around 10 MB a minute in stereo, and the page shows the figure before you start.
Choose the peak mode and leave the target at minus one decibel. Every sample is multiplied by the one figure that puts the loudest moment of your recording exactly there, which is the most the file can take with nothing flattened. The page shows how many decibels that turned out to be.
Clipping is what happens when a sample is pushed past full scale: the top of the waveform is cut off flat, and no later step restores it. Saying a gain "may clip" helps nobody, so the exact number of samples affected is counted and compared with the total, which is what makes the warning worth reading.
Six decibels doubles the amplitude of every sample, but doubling amplitude is not doubling loudness: most listeners describe plus six as roughly half again as loud. The rule of thumb for something sounding twice as loud is about ten decibels, which is a little over three times the amplitude.
Up to a point. Raising the level raises everything equally, including hiss, hum and room noise, so a recording made at a very low level comes back louder but no cleaner. Nothing in a volume control can separate the wanted sound from the noise underneath it, and the page says so when a file is very quiet.
Normalizing finds the loudest sample and scales the whole file so that sample lands on a chosen level, usually one decibel below full scale. Boosting applies whatever change you type, whether or not the file has room for it. Both are the same multiplication; only the way the figure is chosen differs.
Nothing is re-compressed. The browser decodes the file once, every sample is multiplied, and the result is written as an uncompressed WAV, so no second generation of MP3 compression happens. The cost is size: a WAV is roughly ten times an MP3 of the same length.
No upload takes place. Decoding, measuring and multiplying all happen in your own browser, and the finished file is written on your device. No account is needed, no watermark is added to the audio, and the page keeps working with the connection switched off.
Still stuck? Ask about Volume Changer and anybody who knows can answer. No account needed.
Same job, other tools: Frequency Generator – Generate a steady test frequency for speakers, headphones or instruments · Sweep Tone Generator – Sweep smoothly between two frequencies over a time you choose · Audio Sweep Generator – Play a slow logarithmic sweep and hear every rattle in the room.
On the related tools: Mic Test – See your voice on a live waveform and hear it played straight back.
Also works as: Audio Volume Booster · Normalize Audio.
In the collections Audio Tools · Video Tools.