You publish a podcast with episodes recorded on different days, with different microphones, and listening to them back to back the volume jumps from one to the next — the listener has to keep adjusting the volume. "Normalize the volume" sounds like the obvious fix, but peak normalization (what most basic tools do) doesn't actually solve this. You need loudness normalization, and that's where LUFS comes in.
Why normalizing the peak isn't enough
Peak normalization finds the highest sample in the audio and scales the whole thing up (or down) so that peak lands right at 0 dB. The problem: two audio files can have the same maximum peak and sound completely different in perceived volume, because one has an isolated one-second spike while the rest is quiet, and the other stays loud consistently. The peak says nothing about how the ear perceives it over time.
What LUFS is
LUFS (Loudness Units Full Scale) measures perceived loudness, not the instantaneous peak. It's the standard defined by EBU R128, adopted by streaming platforms, TV and podcasts to keep volume feeling consistent across different content. The algorithm (loudnorm in FFmpeg, which is what this tool uses) analyzes the whole audio and adjusts gain so the average loudness hits a specific LUFS target.
The three parameters that control the result
- Target loudness (I, in LUFS): the average perceived volume the audio is adjusted to. More negative means quieter. -14 to -16 LUFS is typical for speech.
- Max true peak (in dBTP): the highest peak allowed after normalization, to avoid clipping (distortion from overload). -1 or -2 dBTP is a safe margin.
- Loudness range (LRA, in LU): how much the volume can vary between the quietest and loudest parts. A low value gives a more compressed, uniform sound; a high value keeps more of the original dynamics.
The three presets, and why they differ
- Podcast (-16 LUFS): the common standard for spoken content, meant to be listened to with headphones or in environments with some ambient noise.
- Music (-14 LUFS): the target used by Spotify and YouTube Music to normalize playback across different songs.
- Broadcast (-23 LUFS): the original EBU R128 standard for TV and radio, with more dynamic headroom because a more controlled listening environment is assumed.
Each preset fills in all three parameters at once, but they stay adjustable — they're a starting point, not a straitjacket.
Frequently asked questions
What's the difference between this and just "turning up the volume"? Turning up the volume (fixed gain or peak normalization) doesn't account for how volume varies throughout the audio. LUFS analyzes perceived loudness across the whole clip and adjusts accordingly.
Which preset should I use for a podcast? -16 LUFS is the standard starting point; set true peak to -1.5 dBTP if you're distributing on platforms that apply their own additional normalization.
Why does my audio sound "flattened" after normalizing? An LRA (loudness range) that's too low compresses the difference between quiet and loud too much. Raise the LRA value if you want to keep more dynamics.
Is it processed locally? Yes, with FFmpeg compiled to WebAssembly running in your browser, with nothing uploaded to any server.
Adjust your audio's volume to the right standard with the audio normalizer: podcast, music and broadcast presets, or manual control over all three LUFS parameters.