Mind Making

To read and edit the audio myself is a huge chore, so I wrote a Python script - txt2wav.py - to read plain text well and record it into a .wav file. Usage: python txt2wav.py book.txt makes book.wav.

I first tried another text-to-speech program, TextAloud, but the result was too poorly phrased to bear. The program didn't even interpret multiple blank lines as a new paragraph. It would just say the words on both sides of the space without any pause between. Stupid.

My script has hundreds of regexps that transform punctuation, abbreviations and spacing into proper pronunciations and phrasing.

The final steps: convert to mp3 using Audacity, add album art using mp3tag, then post to the Internet Archive so I don't have to worry about bandwidth and I don't care above the Creative Commons license requirements.

To work fully, the script needs a Unix dict file. The path to it is specified in the script. The dictionary is used, for amongst other things, a somewhat clever acronym pronouncer that searches all character subsequences of an acronym to find the largest words in it, which seems to be how humans pronounce acronyms, not just spelling out each letter.

All the code is from my Cor machine mind project, which informed my book, Mind Making. Let me know if you find it useful.

The mp3 above was recorded using an old AT&T voice that I'm fond of. Do you know of a better voice that's not too expensive?

Posted at 06:42 PM | Permalink | Comments (0)

Mind Making Bookshelf

I'm starting to keep a list of what I think are the essential books on mind, drawn from philosophy, artificial intelligence and science: Mind Making Bookshelf