argp         posts     research     bugs

Listening to research papers

A friend recently gave me the nice idea of listening to, instead of reading, research papers. We both have “to-read” piles that are constantly getting bigger, even while following the one-paper-per-day rule. Shortly after she mentioned this to me, I started experimenting with Festival, the excellent open source speech synthesis system. I just needed to emerge festival on my Gentoo Linux to get the main engine and emerge mbrola to install some extra natural-sounding voices. I then run echo "(set! voice_default 'voice_us1_mbrola)" >> ~/.festivalrc to change the default male voice to a female one, and echo "Pizza, pizza. Pizza, pizza." | festival --tts - as an initial test. The output was played at double speed. After googling a bit I found the solution to this problem in the Festival FAQ.

To save the output of Festival to an MP3 file I added the following to my
/usr/share/festival/siteinit.scm file:

(Parameter.set 'Audio_Method 'Audio_Command)
(Parameter.set 'Audio_Required_Rate 11025)
(Parameter.set 'Audio_Required_Format 'riff)
(Parameter.set 'Audio_Command "lame --quiet --preset voice $FILE - >> $HOME/tmp/output.mp3")

Next, I simply converted some PDF papers into plain text using pdftotext and fed the output to Festival. Although it is true that not all papers can be fully understood simply by listening to them, this is a way to save significant amounts of time when it comes to not particularly important papers that nonetheless have to be read.