slhck/ffmpeg-normalize
slhck/ffmpeg-normalize has 1,543 GitHub stars and 132 forks. GitHot measured 0 stars in 24 hours, +2 in 7 days, and +3 in 15 days. Last checked 2026-11-20 21:56 UTC. Last push 2026-11-17.
What you can do
- EBU R128 loudness normalization — Two-pass by default, with an option for one-pass dynamic normalization
- RMS-based normalization — Adjust audio to a specific RMS level
- Peak normalization — Adjust audio to a specific peak level
- Selective audio stream normalization — Normalize specific audio streams or only default streams
Gemessener Sternzuwachs
- in 24h
- 0
- in 7 days
- +2
- in 15 days
- +3
- in 1 month
- Verlauf wird gesammelt
Nettoänderungen in überlappenden Zeiträumen, keine Tagessummen. Fehlende Messwerte bleiben leer. Sterne zeigen Interesse, nicht Qualität.
Über uns
Audio Normalization for Python/ffmpeg
Kreativität und MedienREADME.md
Originalffmpeg-normalize
A utility for batch-normalizing audio using ffmpeg.
This program normalizes media files to a certain loudness level using the EBU R128 loudness normalization procedure. It can also perform RMS-based normalization (where the mean is lifted or attenuated), or peak normalization to a certain target level.
✨ Features
- EBU R128 loudness normalization — Two-pass by default, with an option for one-pass dynamic normalization
- RMS-based normalization — Adjust audio to a specific RMS level
- Peak normalization — Adjust audio to a specific peak level
- Selective audio stream normalization — Normalize specific audio streams or only default streams
- Skip files already at target — Avoid re-encoding files already within a threshold of the target level
- Before/after summary — Print a table of input and output levels and the applied gain for all files with
--print-summary - Self-documenting output — Optionally record the effective encoder settings as an
ENCODER_SETTINGStag - Video file support — Process video files while preserving video streams
- Docker support — Run via Docker container
- Python API — Use programmatically in your Python projects
- Shell completions — Available for bash, zsh, and fish
- Album Batch normalization – Process files jointly, preserving relative loudness
🚀 Quick Start
- Install a recent version of ffmpeg and Python 3.10 or higher
- Run
pip3 install ffmpeg-normalizeandffmpeg-normalize /path/to/your/file.mp4, alternatively installuvand runuvx ffmpeg-normalize /path/to/your/file.mp4 - Done! 🎧 (the normalized file will be called
normalized/file.mkv)
🆕 What's New
Version 1.42.0 can optionally write an
ENCODER_SETTINGStag on each normalized audio stream via--write-encoder-settings. The tag holds an ffmpeg-command-equivalent string — the audio codec, bitrate, sample rate and channels, plus the normalization filter including the measured loudnorm values — so a normalized file documents how it was produced. This is a Matroska/WebM convention and has no effect with--metadata-disable.Example:
ffmpeg-normalize input.wav -c:a libopus -b:a 128k --write-encoder-settings -o output.mkvVersion 1.41.0 automatically picks the correct output audio codec for the output container, so you no longer need to specify
-c:a/--audio-codecunless you want to override the default. PCM is chosen for containers that support it; others will use teh default that ffmpeg picks. See the usage guide for details.Version 1.40.0 can optionally skip files that are already at the target level via
--threshold(e.g.--threshold 0.5, disabled by default). Such files are copied through unchanged instead of being re-encoded. The--print-statsoutput now includes a per-filestatus(normalized,skipped, orerror, plus anerrormessage on failure), and the exit code is non-zero if any file failed to process, so a script can tell what happened to each file.Example:
ffmpeg-normalize input.flac -nt peak -t 0 -c:a flac --print-stats -o output.flacVersion 1.39.0 preserves the input bit depth by default when encoding to formats like FLAC, so 16-bit input stays 16-bit without needing
-e "-sample_fmt s16". Use--no-keep-bit-depthto opt out. It also adds--keep-mtimeto copy the input file's modification time to the output, which is useful for preserving when a track was added to a music library.Example:
ffmpeg-normalize input.flac -nt peak -t 0 -c:a flac --keep-mtime -o output.flacVersion 1.38.0 writes the normalized output directly to the destination without using temporary files
Version 1.36.0 introduces presets with
--preset! Save and reuse your favorite normalization configurations for different use cases. Comes with three built-in presets:podcast(AES standard),music(RMS-based batch normalization), andstreaming-video(video content). Create custom presets too!Example:
ffmpeg-normalize input.mp3 --preset podcastapplies the podcast preset (EBU R128, -16 LUFS) to your file. Learn more in the presets guide.
Version 1.35.0 has batch/album normalization with
--batch. It preserves relative loudness between files! Perfect for music albums where you want to shift all tracks by the same amount.Example:
ffmpeg-normalize album/*.flac --batch -nt rms -t -20shifts the entire album so the average RMS is -20 dB, preserving the original relative loudness as mastered.
Version 1.34.0 brings selective audio stream normalization! You can now:
Normalize specific audio streams with
-as/--audio-streams(e.g.,-as 1,2to normalize only streams 1 and 2)Normalize only default audio streams with
--audio-default-only(useful for files with multiple language tracks)Keep other streams unchanged with
--keep-other-audio(copy non-selected streams without normalization)Example:
ffmpeg-normalize input.mkv -as 1 --keep-other-audionormalizes stream 1 and copies all other audio streams unchanged.
Other recent additions:
- Shell completions (v1.31.0) — Tab completion for bash, zsh, and fish shells. See the installation guide for setup instructions.
--lower-onlyoption — Prevent audio from increasing in loudness, only lower it if needed (works with all normalization types).
See the full changelog for all updates.
📓 Documentation
Check out our documentation for more info!
🤝 Contributors
The only reason this project exists in its current form is because @benjaoming's initial PRs. Thanks for everyone's support!