Audio, video & images

Audio Loudness Normalizer

Measure LUFS and true peak, then normalise to Spotify, podcast or EBU R 128 targets.

MP3, WAV, FLAC, Ogg/Opus, M4A, AIFF; max 100 MB and 45 minutes. Processed on our server in a private temporary folder that is deleted afterwards; the result can be downloaded for about 2 hours.

LUFS

Only used with "Custom target".

dBTP

Only used with "Custom target".

LU

Higher keeps more dynamics. If the source range is larger, ffmpeg compresses it.

Try an example: Normalise a test tone to the Spotify target

How it works

Measures integrated loudness (LUFS), true peak (dBTP) and loudness range (LRA) with ffmpeg's EBU R 128 meter, then normalises the file in two passes to a streaming, podcast or broadcast target with a true-peak limit. Shows the measured values before and after, with a player and a download.

  • Pass 1 measures the file with ffmpeg's loudnorm filter (ITU-R BS.1770 / EBU R 128 meter): integrated loudness (LUFS), true peak (dBTP, oversampled), loudness range (LRA) and the gating threshold.
  • Pass 2 feeds the measured values back into loudnorm with linear=true: if the whole file can reach the target with one gain change without exceeding the true-peak limit, only the gain changes (linear). Otherwise loudnorm switches to dynamic mode, which compresses and limits — the result reports which one happened.
  • loudnorm works at 192 kHz internally; the output is resampled back to the original sample rate. The podcast-mono preset also mixes to mono.
  • Pass 3 measures the finished file again, so the "after" values are real measurements of the download, not the targets (for files over 15 minutes the normaliser's own output report is used instead, to stay within the time limit). Files up to 45 minutes are accepted.
  • Spotify (−14 LUFS, true peak below −1 dB) and EBU R 128 (−23 LUFS) are published targets; YouTube ≈ −14 and Apple ≈ −16 LUFS are widely used measured/recommended values, not contractual specs. Podcast −16 LUFS stereo / −19 LUFS mono are common industry recommendations.

Very short files (under about 3 seconds) give less reliable integrated loudness, because the meter gates in 400 ms blocks.

Sources

For scripts and AI agents

  • JSON API: POST https://audioloudnessnormalizer.usefulweb.site/api/v1/calculate (same calculation as this page)
  • Interface description: OpenAPI 3.1 · AI guide: llms.txt
  • MCP server (Streamable HTTP): https://audioloudnessnormalizer.usefulweb.site/mcp · tool name audio_loudness_normalizer
Example request
curl -s -X POST https://audioloudnessnormalizer.usefulweb.site/api/v1/calculate \
  -H 'content-type: application/json' \
  -d '{…see OpenAPI examples…}'

Related tools

Feedback