tooloora

Change Audio Volume — make a file louder or quieter

Raise or lower the level in decibels, normalise to peak, or hit a loudness target in LUFS — locally, no upload.

Runs locally — nothing is uploaded

Drop audio files here or click

MP3, WAV, M4A, OGG or FLAC · one file · stays local

Runs locally in your browser – no upload.

Up to 30 minutes per file – longer recordings exceed the browser's memory.

Change the volume of an audio file

Drop a file and raise or lower the level in decibels, normalise to the peak, or hit a loudness target in LUFS. The before and after values sit right next to the controls. The change happens entirely in your browser, nothing is uploaded. The result is a new file; your original stays untouched.

What does normalising actually do?

The tool finds the loudest peak in the whole file and scales every sample so that this peak sits just below the maximum. The dynamic range is preserved: quiet passages stay proportionally just as quiet. That is different from compression or limiting, where the difference between loud and quiet is deliberately flattened.

How much is +6 dB?

The decibel scale is logarithmic: +6 dB is roughly double the amplitude, −6 dB roughly half. Perceived loudness does not follow directly — a difference of about 10 dB is what people usually describe as "twice as loud". Small corrections of 2 to 3 dB are often enough to line one recording up with others.

Which target loudness is common?

Broadcast has a standard for it: the European Broadcasting Union's recommendation EBU R 128 defines a programme loudness of −23 LUFS, which put an end to the volume jump between programme and commercials. Podcasts usually sit around −16 LUFS, and the large streaming services normalise to roughly −14 LUFS. All three are offered here.

What is the difference between peak level and loudness?

The peak level is the single highest sample in the file. It tells you how much headroom is left before clipping, but almost nothing about how loud something sounds — a single bang can push an otherwise very quiet recording to the limit. Loudness in LUFS measures what the ear perceives over time instead.

The measurement follows ITU-R BS.1770-4: first the signal runs through K-weighting, two filters that model the frequency response of a head and roll off the bass. Then the procedure takes the mean square over 400-millisecond blocks with 75 percent overlap. Finally a two-stage gate applies: everything below −70 LUFS drops out, and after that everything more than 10 LU below the average. This is why a long pause at the start no longer drags the result down — that gate was the decisive improvement over older metering methods.

Honest limits

  • Clipping cannot be undone. If the recording was already clipped at capture, the cut-off peaks are gone; turning it down will not bring them back.
  • Noise gets louder too. Amplification raises everything, including the noise floor of a quiet recording.
  • No true peak, no limiter. What is measured is the sample peak, not the level that can appear between two samples on reconstruction. Anything above the maximum after a boost is hard-clipped; the tool counts those samples and tells you. For mastering, use an audio editor with a real true-peak limiter.
  • No compression. The dynamic range stays as it is — only the whole level moves.
  • Up to 30 minutes per file, because the audio sits decoded in memory — roughly 21 MB per stereo minute at 44.1 kHz.

Frequently asked questions

Is my recording uploaded to change the volume?

No. The level change happens on the decoded audio in your browser; the file stays on your device.

What does normalising do?

It finds the loudest peak in the file and scales everything so that this peak sits just below the maximum. The result is as loud as possible without clipping — the dynamic range stays untouched, quiet passages stay quiet.

How much is +6 dB?

Roughly double the amplitude: the decibel scale is logarithmic, and +6 dB corresponds to a factor of about 2. Perceived loudness rises more slowly than that — a change of about 10 dB is usually described as 'twice as loud'.

Why does the tool warn about clipping?

If amplification pushes samples beyond the maximum value, the peaks are cut off and audibly distort. The tool calculates in advance how much headroom the file has and warns before that happens.

Which target loudness is the right one?

For broadcast, EBU Recommendation R 128 defines a programme loudness target of −23 LUFS. Podcasts commonly sit around −16 LUFS, and the large streaming services normalise to roughly −14 LUFS. The tool offers all three and measures your file first, so you see the value before and after.

How is the loudness measured?

With the procedure from ITU-R BS.1770-4, implemented here in your browser: K-weighting (a head shelf plus a bass-cutting filter), mean square over 400-millisecond blocks with 75 percent overlap, then two-stage gating — everything below −70 LUFS drops out, and afterwards everything more than 10 LU below the average. That is why a passage of silence no longer drags the result down.

Is there a limiter, and is the peak a true peak?

No and no. Measured is the sample peak, not the true peak that can occur between two samples after reconstruction, and there is no limiter smoothing over the peaks. Anything still above the maximum after a boost is hard-clipped, and the tool counts and reports those samples. For a strong loudness boost, listen back before you publish.