MetaData – 3

OK, so metadata injection can’t be much fun for many of you, but it has its uses down the line, especially tracing origins of material and organising things quickly and being able sort and reorganise yourself and your libraries.

What I really like about injecting meta data into files themselves is that it remains part of the file so even if you move it around to different places, the data about the file can remain with it. This is useful for library sounds and very helpful for things you’re sharing with others and finished designs that may end up being broadcast one day.

It’s also a useful way of generating the logs for your submission automatically.

Adobe Audition is the commercial piece of software that has the most comprehensive set of meta data injector possibilities that I’ve seen in relation to audio. However, I’ve not seen ways to batch enter this data using Audition. It seems to be a manual process, something worth avoiding if you can when time is of the essence.

There are a range of very good commercial batch processing tools out there that should allow you to inject meta data into your sounds, but they are often quite expensive, so I’d try the following free programmes and attached scripts before spending cash.

What follows is an approach to batch meta data injection on your audio files. In order to encourage more investigation of loudness standards, what I show also includes a way of injecting loudness data into finished designs. For this I use FFMPEG to do some EBUR128 analysis and injecting the results into the files.

You have three kinds of audio in the second SDM submission:

1 – sound library sounds that have been processed and evolved with digital technology – these are components of [3] below and belong in a well-categorised library.

2 – impulse responses – fantastic capturing of the sonic signatures of spaces, places and things

3 – finished designs – polished, exciting shapes that rock

Each of the above would benefit from some of the BWF meta data information, though perhaps not all of it for every file.

For example, loudness analysis on very short library files is probably very unreliable as the EBUR128 standard is designed for programme/broadcast material. However, other information such as the source files, recording session, aims, technologies and tools used, plus personnel involved is very useful.

Loudness analysis is probably very useful to inject into files that might be broadcast and certainly handy for you to know where you are with these files but less useful in short sound files as the algo won’t work that well with them.

The origins, locations, techniques and people are all integral to the IR process so it’s great to capture that info and pipe into the file directly if you can.

BWFMETAEDIT

This programme is available in a command line interface and as something with a GUI and has been designed for archivists to be able to examine files and inject information into BWF sound files. Download a version for your OS from here:

mediaarea.net/BWFMetaEdit

To follow along further you’ll need the Commandline version (CLI) of this programme and you’ll also need FFMPEG installed.

On windows you can follow instructions for FFMPEG here:

github.com/adaptlearning/adapt_authoring/wiki/Installing-FFmpeg

For OSX users, FFMPEG can be installed effectively with brew:

brew install FFMPEG

When the commandline version of Bwfmetedit and FFMPEG are installed, you’re good to go with the next phase.

 

Make a file full of meta data you want to inject

Save this as a comma separated value list (.csv). Name it something sensible and put it in the folder where the sounds are that you want to posess this information. Grab a the template above from here: metaData.csv

For your sound libraries, you’ll probably make a separate csv file for each folder in the sound library, but there will be much in-common with all the files no-doubt so populate the template with as much core information that all files need before duplicating it and making a different template for each file.

Get my injector script

This script should help you inject meta data into .wav files.


gist.github.com/tinpark/f8bd4ce4d0fb409b964b9ec2a28a3a9e

To use it, download the script and place it somewhere simple, I keep mine in

~/Dropbox/shellScripts/metaInject.sh

but you can also keep it in

/usr/local/bin

When you know where this script lives, you can call it and use it on a folder of sound files.

cd into the directory where the sounds are

cd my/sound/file/directory

When you’re there, you can use the script, but make sure you’ve put the metaData.csv file inside the same directory.

bash ~/Dropbox/shellScripts/metaInject.sh metadata.csv

Setting copyrights

A useful place to setup copyright flags in your file and to work out what you’re offering with your files is to go here:

en.wikipedia.org/wiki/Creative_Commons_license