Sarah Ita Levitan |
HOME | ABOUT ME | MENTOR | PROJECT | JOURNAL | REPORT |
Journal
Week 5 This week I worked on lengthening the sound segments that I have been working with so far. These segments, called IPUs (inter-pausal units) were originally created by running a Praat script that divided the large sound files at pauses. The script defines a pause as a specified duration without speech, and successfully broke the large sound files into smaller, independent phrases. However, many of the IPUs ended up being too short (less than a second). I read a tutorial on Praat scripting, and attempted to adjust the original script so that the resulting IPUs would be a minimum duration, but ran into several obstacles along the way. I then abandoned Praat and wrote a new Python script to concatenate any IPUs that were less than a specified minimum duration (close to two seconds). I used SoX (a command line utility) to perform the actual concatenation of the sound files. The final version of my script took several hours to run (it had to process hundreds of hours of audio!) but once it was finished, the IPUs were ready for use :) [PREVIOUS] [NEXT] |