TECH TIPS
What’s in Your Metadata?
A Shortcut to Renaming Audio Files
by Patrick Cusack
As a sound editorial assistant for over seven years, I have always had a hard time getting out of work at a decent hour. In early February 2004, I spent an entire weekend at work learning about a relatively new means of recording production audio shot on the film set. We had just started sound editorial on Barry Blaustein’s The Ringer. The location audio mixer recorded the entire audio for the film on a Zaxcom Deva digital audio recorder.
Up until this time, I had usually received all location audio on quarter-inch analogue tape or stereo DAT cassettes. Loading audio from either medium into the computer can be a tedious and very time-consuming task. I usually needed a week––if not two––just to load the audio that was required of the scenes that the picture editor had cut. This would all change on The Ringer. In one day, I managed to copy all of the DVD-RAMs to my computer. Unfortunately, one more task remained. I needed to rename all of the audio files that I had just copied. This was no small task, and needless to say, I wanted to make it easier.
When working with production sound at Soundelux, we like all of our file names to feature the sound roll number to which the audio file belongs, as well as the scene and take that the audio file represents. To do this, we mainly use Synchro-Arts Titan, an auto-assembly program. Typically, you must enter these file names by hand, a time-consuming task I wanted to avoid. However, on that long February weekend, I learned about an unseen attribute that is common to practically all production audio files: metadata. I was hoping that I could use this metadata to label all of my production audio files.
Most production audio is recorded to monophonic or polyphonic 16/24-bit Broadcast WAV (waveform audio format) files. The Broadcast WAV file format is a simple extension of the .wav file format that Microsoft specified in the 1990s. WAV files typically consist of “chunks,” or groups of data, that contain attributes important to reading audio files. WAV files must include at least two chunks (they can include more, but these are merely optional). The first is the format chunk. This specifies the bit depth, number of channels, sample rate and other ancillary data for the audio file. The second is the data chunk, which includes all of the sound data for the file. The Broadcast WAV format adds an additional chunk called the broadcast extension chunk, or “bext” (see Figure 1). It was this particular chunk that caught my attention when I was reading about the Broadcast WAV file format because it can contain a wealth of metadata that is useful and informational.
 Figure 1: The six main fields of Broadcast .WAV data. |
As specified by the European Broadcasting Union, the Broadcast WAV extension contains 12 different fields. Manufacturers of field recorders can pack a lot of data into these fields. These fields include a description of the audio file, the manufacturer of the equipment the audio was recorded on, the date and time, and timecode data. Of particular note is the description field. Companies that manufacture location audio recorders like Zaxcom, Fostex, Aaton and Gallery use the description field to include information about attributes such as the scene when the audio was shot, the take for the scene, the tape the audio file belongs to and any relevant track info.
Utilizing this information, I was able to write a small program that could parse this metadata and allow me to rename my audio files. Now, I could load and rename all of the production audio that was shot on set in about a day. This Mac OS X application, which I call BWF Renamer, can be freely grabbed from the public folder of my Mac iDisk. To access it, go to the Mac OS finder, use the Go menu, select iDisk, and choose Other User’s Public Folder. When prompted for the name of the public folder you wish to access, type in patrickcusack. The program is located in the bwf_renamer folder (see Figure 2).
 Figure 2: The main window of BWF Renamer. |
 Figure 3: Dragging files into BWF Renamer's main window. |
To add files to the program, simply drop your location audio WAV files into the main window (see Figure 3). There are eight columns of information displayed in the program. They include the current file name of the audio file, the scene and take derived from the Broadcast WAV metadata, the number of audio channels that the file comprises, the timecode start and stop of the audio file, and, lastly, the new file name that you can use to rename the file. The new file name column is derived from the scene and take columns (see Figure 4).
 Figure 4: A new file name generated from the file's metadata. |
If you wish to add a sound roll number to the beginning of the file, you can enter it in the Sound Roll field in the lower left of the window. This will precede the scene and take. You can also specify a character to serve as the scene and take separator. The program default is “=.” When entering data during a film shoot, some location mixers add additional text in their scene and take fields. If there is anything that you wish to remove, you can enter the offending text in the Remove from Scene or Remove from Take fields. For example, on occasion, I will encounter takes labeled as t 1. To remove this, one would enter t in the Remove from Take field, and this will change the take to now read 1 (see Figure 5). When you are ready to rename your files, press the Rename button, and all of the files will be renamed. For additional information about the program, look at the Read Me file that is included with the program.
 Figure 5: Extraneous text can be removed from either the scene or take. |
As informative as “bext” metadata is, it does have some drawbacks. There are a set number of fields in which you can store data, and those fields are fixed in their lengths. For example, the description field can only contain 256 characters. While this may seem like a lot of space, it might be insufficient to describe a complex sound effect for library purposes, or store track information for eight-track audio recorders. To get around these limitations, a group of audio vendors working with the Institute of Broadcast Sound came up with iXML. iXML is a data chunk just like the Broadcast WAV chunk; however, it can be extended to include new data as the iXML specification changes, while simultaneously preserving backwards compatibility.
There are roughly six broad categories of data in the version 1.5 iXML specification, and they are thoroughly detailed on the iXML website (www.ixml.info). I will touch on a couple of them that were helpful to me. The first is the project information for the audio file. This contains scene and take information as well as any user data for the particular take, such as if the take is a select. Next is the speed category, which contains information such as the timecode rate the audio was shot against, as well as the correct sample rate the file should be played back at. Lastly, the track list category expands on the data that is typically available in the Broadcast WAV’s description field, by detailing the name, function and index information for all of the various channels of audio in the file.
Unfortunately, all of this metadata (either Broadcast WAV or iXML data) is only helpful if the location mixer enters it, but in the past few years, there has only been one show out of the 20-plus with which I’ve been associated that did not feature any useful metadata whatsoever. If you would like to get an idea of all the data that is located in a Broadcast WAV file, you can download another new utility from the bwf_info folder in the iDisk mentioned previously. BWFAssetManager is a slimmed-down version of another program that I use in my day-to-day work (see Figure 6). You can add files to the window by selecting File Menu> Add Files. When you click on a file in the list, any bext metadata contained in the file is displayed on the lower left of the program window, and the iXML metadata is displayed in the middle. You can also audition a file with the controls on the lower right of the program window (see Figure 7).
 Figure 6: The BWF Asset Manager main screen. |
 Figure 7: Adding files to BWFAssetManager. |
I would encourage you to pull down both of these utilities and investigate all of the data that you might be sitting on. You might come across data that will help you or give you clarification on the audio files you already have. And who knows––you might save yourself a weekend otherwise spent renaming more files.
For further information, see www.ixml.info and www.ebu.ch/en/technical/publications/userguides/bwf_user_guide.php.
Patrick Cusack is a sound editorial assistant at Soundelux in Hollywood. He has worked on over 25 films, including Little Miss Sunshine, Borat and The Heartbreak Kid. He recently finished working on The Rocker for Fox Atomic.
[ return to top ]