You upload a video and it sounds quieter than everything else, or so loud that people comment "my ears!". Instead of nudging the volume up and down, matching a LUFS target fixes this cleanly. This guide explains what LUFS is, roughly where each platform sits, and how to match it with the FreeSoundTools Loudness Meter.
LUFS is "how loud it sounds to people"
Loudness used to be judged by the highest point of the waveform (the peak). But two files with the same peak can sound completely different: one dense and full, one with only occasional spikes. LUFS, defined in the international standard ITU-R BS.1770, weights the frequencies our ears are most sensitive to, ignores very quiet parts, and averages the rest into a single "perceived loudness" number. It's negative, and closer to 0 means louder. For example, -14 LUFS sounds much louder than -23 LUFS.
True peak is the "no-distortion line"
Pushing loudness up can make momentary peaks go over 0 dB and distort. Especially after MP3 or AAC compression, the real waveform between digital samples can swing higher than the samples themselves. True peak (dBTP) estimates that in advance. Keeping it below about -1 dBTP means little risk of distortion after compression.
Common targets by platform
- YouTube and music streaming: around -14 LUFS. YouTube turns down videos that are louder than its reference, so making them louder gains nothing.
- Podcasts: around -16 LUFS is common. Comfortable for long listening on earbuds.
- Broadcast: -24 LKFS in the US (ATSC A/85; LKFS is the same unit as LUFS) and -23 LUFS in Europe (EBU R 128).
- Narration and course audio: around -19 LUFS keeps it from jumping out in a quiet room.
Platform targets can change, so double-check each service's upload guidelines for the exact value.
How to match it with FreeSoundTools
- Add your file to the Loudness Meter. It shows integrated LUFS, true peak and loudness range (LRA) right away.
- Pick a target of -14, -16, -19 or -23, or type your own. Set the maximum true peak, for example -1 dBTP.
- Press "Match". The whole file is moved to the target, and a limiter gently catches any moment that would exceed the peak ceiling. If the limiter leaves it short of the target, it re-adjusts automatically.
- Compare the original and the result, then save as WAV, FLAC, MP3 or M4A.
Common mistakes
- If only the voice is too quiet in a video with music, level the voice first with Studio Sound instead of raising the whole mix.
- Files with a very large LRA can leave quiet parts too quiet. Compress (level) them first.
- For a series, match every episode to the same target so viewers don't have to touch the volume each time.
How to read the numbers
Integrated LUFS is the average perceived loudness of the whole file, true peak is the highest moment, and LRA is the difference between quiet and loud parts. For talk content, an LRA of about 5 to 10 LU is easy to listen to; above 15 LU, listeners can miss the quiet parts. The results table shows before and after side by side, but always check with your ears too. When matching a series, write down your target and maximum true peak and use the same values every time.
Open the Loudness Meter (LUFS)
Sources and date checked
Checked: October 6, 2026
- ITU-R BS.1770 — https://www.itu.int/rec/R-REC-BS.1770
- EBU R 128 — https://tech.ebu.ch/publications/r128
- ATSC A/85 — https://www.atsc.org/atsc-documents/a85-techniques-for-establishing-and-maintaining-audio-loudness-for-digital-television/
Frequently asked questions
What is the difference between LUFS and dB?
dB (peak) describes the highest point of the waveform, while LUFS describes the average loudness people actually hear. LUFS is the natural choice for matching loudness.
What LUFS should I use for YouTube?
Around -14 LUFS works well. YouTube turns down videos that are louder, so there is no need to go higher.
Do I have to stay under -1 dBTP?
It is not mandatory, but staying below -1 dBTP is a safe way to avoid distortion when the file is compressed to MP3 or AAC.
My numbers differ slightly from another program.
Differences in measured range, silence handling and sample-rate conversion can cause 0.1 to 0.5 LU of variation. If the difference is large, check that the file loaded correctly.