CHENZI XU

This Post has been updated and migrated to here.

​

Acoustic analysis programmes such as Praat, ESPS, and Matlab have enabled the large-scale processing and analysis of speech data. For phoneticians, we often work with a corpus of audio files and begin with the preparation of the speech data involving reformatting, transcription, and segmentation. Such pre-processing steps made it convenient for us to focus, retrieve, and extract the target units of our research. Many tools with strong computational power for relatively big data are available now, which automates some processes and greatly boost our efficiency.

 

In the course of my DPhil, I have accumulated a bunch of scripts that greatly speeded up some processing procedures. Some are directly downloaded online from those who kindly shared their efforts, others are either adapted from others to suit my own purposes or my nascent attempts in a few programming languages. Dipping my toes in scripting was a steep learning curve but worthwhile. Here I would also like to share some useful scripts I had (and brief tutorials of how to use them), to pay it forward to the community where their generous sharing helped me a lot. This also serves as an archive of some of my analyses procedure, making the research open and reproducible.

​

Prerequisites:

• Bash/Unix Shell Basics in the Terminal (A good tutorial here)

• Basic familiarity with Praat.

• Googling or stackoverflowing your problems!

​

Topics covered so far:

01 Forced Alignment (#P2FA and Montreal #Python #Mandarin examples)

02 Polynomial Pitch Curve Fitting (#Matlab #ESPS)

​