Home

Awesome

ffmpeg-normalize

Build Status PyPI version

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.

Batch processing of several input files is possible, including video files.

Contents:

Requirements

Installation

pip3 install ffmpeg-normalize

Or download this repository, then run pip install ..

Usage

ffmpeg-normalize [-h] [-o OUTPUT [OUTPUT ...]] [-of OUTPUT_FOLDER] [-f]
                    [-d] [-v] [-n] [-pr] [--version] [-nt {ebu,rms,peak}]
                    [-t TARGET_LEVEL] [-p] [-lrt LOUDNESS_RANGE_TARGET]
                    [-tp TRUE_PEAK] [--offset OFFSET] [--dual-mono]
                    [-c:a AUDIO_CODEC] [-b:a AUDIO_BITRATE]
                    [-ar SAMPLE_RATE] [-koa] [-prf PRE_FILTER]
                    [-pof POST_FILTER] [-vn] [-c:v VIDEO_CODEC] [-sn]
                    [-mn] [-cn] [-e EXTRA_OUTPUT_OPTIONS]
                    [-ofmt OUTPUT_FORMAT] [-ext EXTENSION]
                    input [input ...]

For more information, run ffmpeg-normalize -h, or read on.

Description

Please read this section for a high level introduction.

What does the program do?

The program takes one or more input files and, by default, writes them to a folder called normalized, using an .mkv container. All audio streams will be normalized so that they have the same (perceived) volume.

How do I specify the input?

Just give the program one or more input files as arguments. It works with most media files.

How do I specify the output?

You can specify one output file name for each input file with the -o option. In this case, the container format (e.g. .wav) will be inferred from the file name extension that you've given.

If you don't specify the output file name for an input file, the container format will be MKV, and the output will be written to normalized/<input>.mkv.

Using the -ext option, you can supply a different output extension common to all output files, e.g. -ext m4a.

What will get normalized?

By default, all streams from the input file will be written to the output file. For example, if your input is a video with two language tracks and a subtitle track, both audio tracks will be normalized independently. The video and subtitle tracks will be copied over to the output file.

How will the normalization be done?

The normalization will be performed with the loudnorm filter from FFmpeg, which was originally written by Kyle Swanson. It will bring the audio to a specified target level. This ensures that multiple files normalized with this filter will have the same perceived loudness.

What codec is chosen?

The default audio encoding method is uncompressed PCM (pcm_s16le) to avoid introducing compression artifacts. This will result in a much higher bitrate than you might want, for example if your input files are MP3s.

Some containers (like MP4) also cannot handle PCM audio. If you want to use such containers and/or keep the file size down, use -c:a and specify an audio codec (e.g., -c:a aac for ffmpeg's built-in AAC encoder).

Examples

Normalize two WAV files and write them to the specified output files with uncompressed PCM WAV as audio codec:

ffmpeg-normalize file1.wav file2.wav -o file1-normalized.wav -o file2-normalized.wav

Normalize a number of videos in the current folder and write them to a folder called normalized, converting all audio streams to AAC with 192 kBit/s.

ffmpeg-normalize *.mkv -c:a aac -b:a 192k

For Windows, the above would be written as a loop:

for %%f in ("*.mkv") do ffmpeg-normalize "%%f" -c:a aac -b:a 192k

Normalize an MP3 file and write an MP3 file (you have to explicitly specify the encoder):

ffmpeg-normalize input.mp3 -c:a libmp3lame -b:a 320k -o output.mp3

Normalize many files, keeping PCM audio, but choosing a different container:

ffmpeg-normalize *.wav -c:a pcm_s16le -ext aif

Instead of EBU R128, one might just want to use simple peak normalization to 0 dB:

ffmpeg-normalize test.wav --normalization-type peak --target-level 0 --output normalized.wav
ffmpeg-normalize test.wav -nt peak -t 0 -o normalized.wav

You can (if you really need to!) also overwrite your input file. Warning, this will destroy data:

ffmpeg-normalize input.mp4 -o input.mp4 -f

If you need some fancy extra options, such as setting vbr for the libfdk_aac encoder, pass them to the -e/--extra-options argument:

ffmpeg-normalize input.m4a -c:a libfdk_aac -e='-vbr 3' -o output.m4a

Further examples? Please submit a PR so I can collect them.

Detailed Options

File Input/Output

General

Normalization

EBU R128 Normalization

Audio Encoding

Other Encoding Options

Output Format

Environment Variables

The program additionally respects environment variables:

FAQ

The program doesn't work because the "loudnorm" filter can't be found

Make sure you run ffmpeg v3.1 or higher and that loudnorm is part of the output when you run ffmpeg -filters. Many distributions package outdated ffmpeg 2.x versions, or (even worse), Libav's ffmpeg disguising as a real ffmpeg from the FFmpeg project.

Some ffmpeg builds also do not have the loudnorm filter enabled.

You can always download a static build from their website and use that.

If you have to use an outdated ffmpeg version, you can only use rms or peak as normalization types, but I can't promise that the program will work correctly.

Should I use this to normalize my music collection?

When you run ffmpeg-normalize and re-encode files with MP3 or AAC, you will inevitably introduce generation loss. Therefore, I do not recommend running this on your precious music collection, unless you have a backup of the originals or accept potential quality reduction. If you just want to normalize the subjective volume of the files without changing the actual content, consider using MP3Gain and aacgain.

Why are my output files MKV?

I chose MKV as a default output container since it handles almost every possible combination of audio, video, and subtitle codecs. If you know which audio/video codec you want, and which container is supported, use the output options to specify the encoder and output file name manually.

The conversion does not work and I get a cryptic ffmpeg error!

One possible reason is that the input file contains some streams that cannot be mapped to the output file. Examples:

The default output container is .mkv as it will support most input stream types. If you want a different output container, make sure that it supports your input file's video, audio, and subtitle streams (if any).

Also, if there is some other broken metadata, you can try to disable copying over of metadata with -mn.

What are the different normalization algorithms?

Couldn't I just run loudnorm with ffmpeg?

You absolutely can. However, you can get better accuracy and linear normalization with two passes of the filter. Since ffmpeg does not allow you to automatically run these two passes, you have to do it yourself and parse the output values from the first run. If this program is too over-engineered for you, you could also use an approach such as featured in this Ruby script that performs the two loudnorm passes.

After updating, this program does not work as expected anymore!

You are probably using a 0.x version of this program. There are significant changes to the command line arguments and inner workings of this program, so please adapt your scripts to the new one. Those changes were necessary to address a few issues that kept piling up; leaving the program as-is would have made it hard to extend it. You can continue using the old version (find it under Releases on GitHub or request the specific version from PyPi), but it will not be supported anymore.

Can I buy you a beer / coffee / random drink?

If you found this program useful and feel like giving back, feel free to send a donation via PayPal.

License

The MIT License (MIT)

Copyright (c) 2015-2018 Werner Robitza

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.