Creating a .aup with xslt What is the best practice?

I am looking to streamline my dictation process by using xforms to generate the boilerplate portions of documents, but I need to preserve the ability to dictate a sound file with the more free form parts to be typed in by my secratary. I am good at xml/xslt, but not an audiophile by any stretch. Before I start hacking in earnest I think I’d like to get a little feedback from the pros.

I propose sending my transcriptionist a zipped archive containing an aup file the projname (_data) directory and an xml file containing boilerplate elements interspersed with dictation placeholder elements. (When I get really cool, I might wrap it all up gor open office.)

I can experiment to find the best values for some settings (unless someone has suggestions for mono voice optimized for file size).

I will need to make a my_aup.aup file which references the dictation placeholders found in the xml document at the same time as I create the boilerplate elements. Then I’ll open this with audacity and fill in the voice. My first thought was to use a label track. That seems wrong now because the labels couldn’t be put in the right place unless I knew how long each dictation was. Looking at the xml layout and the DTD shows that labeltrack timetrack and wavetrack have name attributes which I could populate with the name of the dictation placeholder. I’m going to guess that I want a on-to-one relationship between dictation placeholders and wavetracks. Can somebody verify that.

I also suspect that filename=“” is the binary audio file that is modified when I edit the .aup using audacity. Some questions: do I need to create this file ( ie touch on the disk or will audacity do that when I start recording? Similarly, the DTD shows a bunch of stuff (suspiciously some has ‘legacy’ in the name) as required, but when I cntrl+shift+N to create a new track, only

is created. Is that really all I need for each placeholder?

Here is where my non-audiophilia comes in. These wavetracks don’t overlap do they? When I think of recording tracks I think about the “wall of sound” on the ‘Let it Be’ album where all the tracks play over each other. I’m not looking for that.

BTW. Thanks for a top notch app that really make great use of the spirit of xml.

To follow up my own thread in case anybody has a similar project.

In fact my first instinct to use labels was probably best. I thought that the reason I couldn’t record voice over an empty track was a hardware problem with my system. Turns out to be a basic safety rule in audacity. And a sensible one at that. So my new workflow is:

xforms → xml (with boilerplate and placeholders) → xslt → aup (with one label for each placeholder) → open with audacity and record a single sound track then slide the labels over the relevant parts of the track. → zip into an archive → send for transcription

The xslt for this is now really easy. I should have working examples this weekend if anybody has a similar project. But please pipe in if you think I’m going down the wrong track. To bad I can’t just select a track and record over it.