SLAM : a method for the automatic Stylization and LAbelling of speech Melody (optimized only for LINUX distributions) Compatible with Python 3.6
N. Obin, J. Beliao, C., Veaux, A. Lacheret (2014). SLAM: Automatic Stylization and Labelling of Speech Melody. Speech Prosody, 246-250.
-
Download or clone SLAM and swipe-installer and put them in the same repository.
-
Install SWIPE module
Swipe, by Kyle Gorman (http://ling.upenn.edu/~kgorman/c/swipe/), is a pitch estimation algorithm which is required for SLAM to work. Sources of swipe are provided in the swipe-installer directory. These are slightly modified versions different from the official github release. Modifications are only for the purpose of swipe compiling under C89 instead of C99 standards.
-
Install the following libraries required by SLAM:
sudo apt-get install python-numpy python-scipy python-matplotlib python-pandas python-sympy python-nose python-chardet
-
Drop your wav files and textgrid files in the corresponding directories. wav and textgrid files must come in pair of the same name example: "myfile1.wav" "myfile1.TextGrid" "myfile2.wav" "myfile2.TextGrid"
-
Open a terminal and go to the SLAM directory
-
Execute
python SLAM.py -
Follow the instructions.
you can open SLAM.py and modify the parameters to suit your needs.