An Automator workflow for extracting text as speech from PDF files

While Automator is pretty useful, I personally find the Terminal / command line a nice place to solve problems like this.

The basic idea is still using the steps you describe, but doing all the work from the command line. I researched a useful PDF-to-Text converter, and I found PDFminer quite useful. If you can get it to run, half of your work is done!

pip install pdfminer

Step one and two, then can be solved with this one-liner in Terminal:

pdf2txt.py example.pdf | say -v Daniel -o example.aiff

Still missing is the addition of metadata — what do you need here: Title / Album / "Artist"?

In a final step, you would add the file to a certain iTunes playlist. Depending on your ideal workflow, one then could build a little LaunchAgent that monitors a folder for new files...