Audio_conv.py
 -Overview
 -Requirements
 -Download
 -Usage
Changes
Known Problems

---

Audio_conv.py

Overview

This script was created as a simple tool for converting between several different audio formats under Linux, Unix and Windows (OS X should work, but not tested). Conversion works by converting the input file(s) to an intermediate wav file, optionally normalizing the wav, then writing out the output file with appropriate tag information.

There are lots of other scripts that do similar things, but I created this script to handle wildcards, work recursively, and handle filenames with spaces and other strange characters (question marks, single quotes, double quotes, etc). Hopefully this script saves you some time with file conversion, it should be especialy useful for making mp3 CDs for portable mp3 players. It may look complicated, but its actually quite simple to use; just look at some of the examples and play around with it a bit. Its released under the GNU GPL version 2.

Here's a few other OSS audio manipulation programs that I've found but haven't used:

»http://soundconverter.berlios.de/ GTK and Python based GUI conversion program that uses Gstreamer for conversion of audio formats, very impressive program. Great for those who prefer GUIs. Currently it cannot read wma files.

»http://sox.sourceforge.net/

»http://www.mbox.wz.cz/

»http://pacpl.sourceforge.net/

»https://savannah.nongnu.org/projects/audio-convert

»http://klira.sourceforge.net/

Mp3gain »http://mp3gain.sourceforge.net/ is a great program that can adjust the gain of mp3s with a lossless method that doesn't rely on replayGain settings. This is only useful for mp3s however, so its not used by this script. This script uses normalize to normalize the intermediate .wav file before converting to the desired output format.

Requirements

You must have at least some of the following installed, depending on what you're using the script for (eg: if you're not reading or writing Flac files, you dont need Flac installed). The versions listed were used for testing, but any version with the same functionality should work. All of these programs are available in Debian or Ubuntu Linux. This script has been run under windows, but each dependency must be downloaded and installed (all are available for windows and *nix).

• »Python 2.4 or greater (using the subprocess module).
• »Mplayer to decode wma files (version 1.0pre5-3.3.4).
• »vorbis-tools to decode and encode ogg vorbis files (version 1.0.1).
• »lame to encode/decode mp3s (version 3.96.1).
• »mp3info for mp3 tags (version 0.8.4)
• »flac for flac files (version 1.1.1)
• »normalize for normalization (version 0.7.6)

Download

• current version (0.4.5.3): »upload/audio_conv.py

Usage

• Download the file »upload/audio_conv.py, and make it executible (eg: "chmod +x audio_conv.py"). It is easiest if all required programs are included in the path.

• Run the script as follows:

 "./audio_conv.py"
 or:
 "python audio_conv.py"
 "./audio_conv.py -h" will display detailed help info and usage.

• If you want to be able to run the program from any directory, put it in your path. For example, an easy way to do this is to copy the file to /usr/bin when logged in as root (make sure its executable).

• The following is the usage for the program:

usage: audio_conv.py -i <input> [options] --to-mp3 | --to-ogg | --to-wav | --to-flac

        This script is for conveting from an input audio format to an
        intermediate wav file, then outputting to the selected output
        format while keeping the tag info.  At first this sounds crazy,
        why go from one lossy format to another?  This may be the only
        option for your portable mp3 player.  It is also an easy way to
        convert a music collection from a high quality format (eg: flac
        or wav) to a different format (eg: mp3 or ogg vorbis).

        Optionally works recursively, keeps the tag information, and handles
        filenames with strange characters such as spaces, single quotes, and
        question marks.  It can also output the converted file to a different
        directory, maintaining the directory structure. Most of the other
        wma2ogg or ogg2mp3 scripts fail on these or dont do recursion, which
        is why this script exists.

        Supported input formats:
                wma
                flac
                wav
                ogg
                mp3
                rpm RealAudio streams
                rm/ra RealAudio files

        Supported output formats:
                wav
                flac
                mp3
                ogg


        Requires mplayer (for wma), vorbis tools, flac, lame, normalize
        and mp3info.  Note: if there are no tags in the input file, the
        filename will be used as the 'title' tag in the output file.

        Examples:

        This example converts the ogg vorbis file 'foo.ogg' to 'foo.mp3':
         audio_conv.py -i foo.ogg --to-mp3

        This example converts all wma files to ogg vorbis files in a
        directory (not recursive and wont delete the original files):
         audio_conv.py -i "*.wma" --to-ogg

        This command passes a string to the encoder (lame in this case,
        since mp3s are being created), which can be any sort of option
        (variable bitrate, mode, etc).  The option here specifies that
        variable bitrate level 6 is to be used (see encoder documentation
        for more info):
         audio_conv.py -i file.mp3 --to-mp3 --encoder-option "-V 6"

        This dry-run example lists all the files that would be processed,
        but are simply listed instead (an output format is needed, so
        --to-mp3 is chosen though no conversion is done):
         audio_conv.py -i "*.wma" -r --to-mp3 --dry-run

        This command converts all recoginized files to mp3s recursively,
        performs normalization and deletes the original files.  If the
        file doesn't have a recognized extension, then it is skipped.
         audio_conv.py -i "*.*" --to-mp3 --recursive --normalize --delete

        Here is the same command with short options:
         audio_conv.py -i "*.*" --to-mp3 -r -n -d

        This is the same command, but with a different destination directory.
        This will duplicate the directory structure present in the source
        to subdirectories of the destination directory populating it only with
        the converted files.
         audio_conv.py -i "*.*" --to-mp3 -r -n -d --dest-dir /home/user/newdir


options:
  -h, --help            show this help message and exit
  -i INPUT, --input=INPUT
                        Specify input file, accepts wildcards in quotes.
  -b BITRATE, --bitrate=BITRATE
                        Specify bitrate of output file.  Some encoders(eg
                        Lame) will only accept certain values such as 32, 40,
                        48, 56, 64, 80, 96, 112, 128, etc. The default bitrate
                        of the output file is the same as the input bitrate.
                        This option will be ignored if the '--encoder-option'
                        is used.
  -r, --recursive       Convert all files matching the input filename in
                        current directory and all subdirectories.  If using
                        wildcards, they must be in quotes.
  --dry-run             List the path of each input file and output file but
                        do not convert anything.
  --to-ogg              Ogg vorbis output format.
  --to-mp3              Mp3 output format
  --to-wav              Wav output format
  --to-flac             Flac output format
  -n, --normalize       Normalize the volume of output files.
  -d, --delete          Delete the input file after conversion (use with
                        caution!).
  -f, --force           Force conversion of files even if the input and output
                        bitrate are the same [default skips these files].
  --dest-dir=DESTINATION
                        Specify a different output directory. This will
                        duplicate the same file and directory structure seen
                        in the source directory, and populate it with
                        converted files. If the destination directory does not
                        exist, it will be created. Works especially well with
                        the recursive option.
  -t PATH, --tempfile=PATH
                        Specify the path for a user defined tempfile. The
                        default tempfile uses a system tempfile.
  -e STRING, --encoder-option=STRING
                        Specify options to be passed to the encoder. May
                        include any option supported by the encoder such as
                        variable bitrate encoding. This option overrides the
                        'bitrate' option, so a custom bitrate or vbr setting
                        should be specified otherwsie the encoder will use its
                        default bitrate. Be careful not to specify tags that
                        may conflict with tags taken from input file. String
                        must be encapsulated by quotes.
  -v, --verbose         Show all standard output and error messages from
                        backend programs.  Default is to hide these messages.

---

Changes

• 0.1 Uses a system tempfile instead of a file named "tempfile", so multiple instances can be running at the same time.

• 0.2 Redesign for cleaner code, and slightly simpler use. Added flac and wav as input and output formats

• 0.3 Ported to windows (with downloaded exe's), and a newer version of mplayer that uses a different syntax.

• 0.3.1 Jim Hanley added support for decoding realaudio .rpm streams and rm/ra files.

• 0.3.2 Added a strict filter to remove genre tags that are not in a 'genre list', should solve the problem of Lame (or other encoders) failing to encode due to crazy genre tags.

• 0.3.3 Made default bitrate of output file equal to the input bitrate unless the bitrate option is specified. Skipping files that have a bitrate less than or equal to the specified output bitrate, this can be bypassed with the --force option.

• 0.4 Added support for Metaflac tags, and the option of using a different output (destination) directory as requested by Joe Oldak.

• 0.4.1 Small bugfix for decoding of files with Mplayer.

• 0.4.2
• added -nolirc -nojoystick and to the mplayer decoder commands to reduce the warnings printed to standard output.
• Changed the creation of a tempfile when decoding with mplayer. Unix uses a normal tempfile, while 'nt' systems use a tempfile in the local directory.
• There is a new --dry-run option that lists the files to be converted
• There is also a --encoder-option option that allows passing of options to the encoder (currently lame, oggenc and flac). This is useful for things like vbr that are hard to make universal to every encoder.

• 0.4.3 Added the --tempfile option to use a user specified tempfile for the pcm file.

• 0.4.4 Feb 2006: refactored the recursion logic to make it more robust and avoid known bugs with Glob. This should fix most problems with recursion.

• 0.4.5 Added an extra check so the original input file won't be deleted if the output file does not exist (when using the --delete option).

• 0.4.5.1 Small bugfix for reading of tags from a Flac file (and small cleanup), so filenames with spaces will be handled normally (Thanks to J. Huckabay).

• 0.4.5.2 Improved the "--dry-run" option; it will now show the path of the input and output file even if a different destination directory is chosen. This is very handy for seeing what will be converted prior to running the script. Also switched from using os.system() to using subprocess.Popen() because control-c works much better among other things. Refactored other try statements so control-c won't mess them up. Added the ability to handle files with double quotes in the filename (crazy, but they exist). These can exist in *nix, but windows won't allow them - so replacing " with \" works everywhere.

• 0.4.5.3 small fix to replace double quotes in meta tags for flac/ogg/mp3 with single quotes. This will allow songs with two double quotes in the tags to be converted properly.

Known Problems

• If running in windows, and the executables for the other programs (lame, mplayer, etc) aren't in the path, these can be set manually near the top of the script (a commented example is included).

• Normalize in debian/ubuntu is now called normalize-audio. You may have to change the name of this to conform to your system.

• Mplayer has changed the syntax for pcm (wav) output. The pcm -aofile <filename> options has changed to -ao pcm:file=<filename> which doesn't like dos filenames (c:\bla\...). so I'm using a real hack to make a tempfile in the current directory (for windows only, *nix works normally). This script uses the syntax for the newer version of Mplayer, if you need the older syntax comment out the current method near line 286 and uncomment the one above it. This should be pretty clear when looking at the code.

• Characters such as å and ö will be converted correctly if the terminal supports it. However, other characters such as asian ones may be converted to question marks.

Email: crleblanc at gmail dot com.