You finish a podcast episode, play it next to another show, and yours is noticeably quieter. You raise the volume until the loudest word reaches the top, and it is still quieter. The reason is that volume as you hear it is not the same as the highest peak in the file. This guide explains the measure that streaming services use, LUFS, and how to choose a target in Normalize Audio.
Peak level and loudness are different
A peak meter shows the single highest moment of the waveform. One loud clap or a plosive "p" can sit at the top while the rest of the recording is quiet. Loudness is how strong the sound feels over time. LUFS (Loudness Units relative to Full Scale) is a measure of that perceived loudness, defined in standards ITU-R BS.1770 and EBU R128. When the tool shows "how loud each file is now", it is using this measure, not the peak.
Why this matters: two files that both peak at the same point can differ by many LUFS, and listeners hear the difference. Platforms that level audio to a target will turn down a file that is louder than their target, so making a file louder than the platform wants gains you nothing.
The three presets and who they are for
| Preset | Typical use | Comment |
|---|---|---|
| −14 Streaming | Music and video for streaming platforms | Common reference for many streaming services; leaves some dynamic range. |
| −16 Podcast | Spoken-word episodes | A bit quieter than music, which suits speech listened to in the car or on earbuds. |
| −23 Broadcast | European television and radio delivery | The EBU R128 reference; much quieter than streaming. |
| Custom | Anything else | The slider runs from −40 to −5 LUFS in steps of 0.5. |
Services change their rules, so check the current loudness guidance of the platform you publish on if a specific number matters. When you do not know the destination, −16 for speech and −14 for music is a reasonable choice.
How to normalize several files
- Drop one or many MP3, WAV, OGG, M4A or FLAC files. Each is measured as soon as it is added.
- Choose the target. The list shows how many dB each file will change: for example a badge such as +4.2 dB for a quiet file or −3.0 dB for a loud one.
- Keep Prevent clipping on. When a gain would push a peak over the ceiling, a short look-ahead limiter holds the peaks at −1 dBFS. A Peaks limited badge tells you where this happens; with the switch off the badge reads May clip.
- Save as MP3 192k or WAV. Several files download together as a ZIP.
The trade-off: loud target, limited peaks
Raising a dynamic recording to a loud target means the peaks would exceed full scale. The limiter prevents distortion by pressing those peaks down, which reduces the difference between loud and quiet moments. For speech this is barely audible. For music with big dynamics, such as classical or film scores, a quieter target like −16 or −23 keeps more of the original feel. If a file shows Peaks limited and you care about transients, try a lower target and compare.
Normalizing is not boosting
The Volume Booster multiplies the signal by a number you choose, up to 500%. It does not look at the content. Normalize Audio measures first and applies exactly the gain needed. Use the booster when you want a specific, known change; use normalization when you want several files to land at the same loudness. Loudness measures also do not change the content inside a file: a podcast with a loud intro and a quiet interview is levelled as a whole, not section by section.
Preparing the files first
- Trim silence and mistakes with the Audio Cutter so they do not influence the measurement.
- Remove steady noise first with Remove Background Noise. Otherwise noise is counted as part of the loudness and the voice ends up quieter than intended.
- Normalize the parts, then join them with the Audio Joiner.
Problems and fixes
- The file shows a gain of 0.0 dB. It already sits at the target; nothing changes.
- The result sounds squashed. The gain was large and the limiter was working. Pick a quieter target or turn the music down at the source.
- A very quiet file is still noisy after normalizing. Raising the level raises the noise by the same amount. Clean first.
- A file is rejected. Use files your browser can decode; MP3, WAV, OGG, M4A and FLAC are the usual ones.
A worked example
Say you edit a weekly podcast. Episode 41 measures quieter than your target and the tool shows a positive gain for it; episode 42, recorded with a hotter microphone setting, shows a negative gain. Both are set to −16 LUFS, so after processing a listener who moves from one episode to the next does not touch the volume. Keep the same target for the whole series: changing it between episodes brings the problem back. Write the target in your production notes, together with the format you saved, so that a future episode, or a colleague, follows the same rule.
For music albums the same idea holds with a twist: keep the relative loudness between tracks that the artist intended. Normalizing every track to one number removes the difference between a quiet ballad and a loud opener, which may not be what you want. In that case, process only the tracks that are clearly out of line.
Frequently asked questions
What does a negative LUFS number mean?
The scale is relative to digital full scale, so everything is below zero. A number closer to zero, such as −14, is louder than −23.
Can normalizing make a bad recording good?
No. It changes the level only. Noise, echo and distortion stay.
Are my files uploaded?
No. Measuring and processing happen in your browser.
Which format should I save in?
WAV if you will edit further, MP3 at 192 kbps for final delivery.