I will be exploring epub and speech-to-text modules to help automate tests of correctness for the audioreader prototype.

Project Update: Comparing Epubs and MP3s

The audioreader prototype has not changed greatly since the last post. The current issue is dealing with program correctness. The only way I can validate program output is by flipping through a copy of the book I am interested in, but this can be automated somehow.

I am thinking of exploring speech-to-text conversion in Python, though I currently know nothing about it. If it goes as planned, I will have several ideas for improving the audioreader in the near future. Of course, for any of this to be automated, I also need a means to read text from .epub files..

• Find means to scan and parse .epub data
• Transcribe current audio file or listen to narrator’s playback (former more portable) using Sphinx or Google Speech
• Treat epub data as contiguous blocks to minimize time searching for a match on the text (also lowers risk of false positives)