Port of mp3gain to ANSI C/C++, wavgain

Release: 1.3.1/3a (based on mp3gain 1.3.1)

introduction usage download history changes todo contact

Introduction

Mp3gain is a powerful utility to modify the real loudness of MPEG audio files automatically to one common level. It utilizes the psychoacoustic modell of replaygain to do the loudness analysis. The modifications to the MPEG files are lossless (i.e. without any additional losses) and reversible. See the original homepages for further details.

The modification of the MPEG files is done in place and lossless.
The utility can operate on MPEG 1, MPEG 2 and MPEG 2.5 Layer 1, 2 and 3 files.

WAVgain does the same on RIFF wave files.
Due to the nature of the PCM files this procedure is in contrast to MPEG audio files neither lossless nor reversible. However, the losses are small.

Usage

installation of the EMX binaries (OS/2)

Copy the executable files somewhere to your harddrive. You propably want to use a folder which is listed in your PATH environment. Youl must have the EMX 0.9d runtime installed (search at hobbes).
The included gcc321.dll is shared by other applications like mozilla too. You only need one copy in your libpath.

instruction for other platforms

Compile your own binaries using gcc or a gcc-compatible compiler like cygwin. The source should not contain any platform specific stuff. You can use the included Makefile. However, it is most likely that you have to modify the first few lines to match your environment and platform.

usage

Use mp3gain to adjust the loudness of MPEG audio files and wavgain for wave files.

See command line help for details:

mp3gain /? or
wavgain /? respectively

Examples:

To adjust one file with manual control use something like

mp3gain /a /Ia /d2 filename
/a and /r are the same for one file. /d2 tweaks the very "safe" default setting by +2dB. A good choice for pop music.

To adjust a whole album with manual control

mp3gain /a /Ia /d2 track1_file track2_file track3_file ...

On platforms where the command shell expands wildcards (like Unix) you may simply use the following statement to treat all files in the current directory as one album

mp3gain /a /Ia /d2 *.mp3

If you have to process a very large number of files you may not like the interactive mode. However, you will need a shell script for the iteration over the files.

mp3gain /a /Ca /d2 current_file

Download

This program is distributed under the terms of the GNU LESSER GENERAL PUBLIC LICENSE.
ZIP file with C source, OS/2 binaries and this guide: Version 1.3.1/3
The source code should compile on almost any ANSI compliant C/C++ compiler.

History

Version 1.3.1/3a (Jan. 2, 2004)

Some fixes to the layer II support.
Turned the analysis logic into OO code, too.
Added some features of mp3gain 1.4.3. There will be a synchronized realease in the future.

Version 1.3.1/3 (Dec. 16, 2003)

Added full support for layer I and II audio files. Layer I is untested since I do not have any layer I file. Let me know wether it works.
Code cleanup. Now most of the logic is OO.

Version 1.3.1/2a (Dec. 07, 2002)

Added option /i to enable interactive mode and more options in case of clipping.
Fixed /d in wavgain.

Version 1.3.1/2 (Dec. 04, 2002)

Updated port to mp3gain release 1.3.1. First of all, this removed a serious bug in the chaingeGain function which caused broken bit streams.
Option /p is now working.
Added support for 16 bit wave audio files: wavgain.

Version 1.0.5 (pre) first public release

Changes to the original distribution

I have changed the RMS time window functionality completely. The data is now processed by half overlapping windows of 100ms length, scaled by a Hamming window function. This results in an effective window length of around 60ms and reduced systematic errors at low frequencies.
gain_analysis.c and mp3gain.c are now renamed to .cpp. This is because the .c version already used C++ syntax extensions. New C/C++ compilers often do not claim about that, but that is not guaranteed.
The code in gain_analysis.c is optimized significantly. However, some compilers did most of this optimizations themself, gcc 3.0 not. This is ported back to the original distribution.
All Win32 API calls are replaced by their ANSI C pendants, as far as possible.
If a clipping warning is generated while applying album gain either none or all files are processed. This is different from the original implementation where the clipping check is done on per file basis. This was conflicting with the idea of album gain.
Added option /i to enable interactive mode and more options in case of clipping.
Support for Layer I and II audio files.
Disabled the /m option because it's result is no longer predictable due to differences between layer III and layer I & II files.
Complete rewrite of the stream processing state machine. The code is now mainly object oriented and split into several files.

ToDo

Since the original mp3gain has gone a completely different way than this release it is neccessary to merge the code lines to a new release, combining the features of both branches. This has begun with release 1.3.1/3a.
Include a scale factor statistic into the analysis of layer I and II files to detect over- and underflows before the files get modified (not reversible in this case).
Apply the recoding to wavgain, too.
Better wave audio support. First of all: mono files.

Contact

Suggestions, help, complaints (but not too much:-): mueller@maazl.de

Original location of this page: http://www.maazl.de/project/mp3/mp3gain.html