How does the script work?

Here you see a PRAAT window together with the TextGrid file that was produced by the script. The marks 95 to 103 are the syllable nuclei as detected by the script. The yellow line in the speech signal represent intensity (dB), and the blue line voicedness. The script first considers all peaks above a certain threshold (determined with respect to the median intensity of the speech file) as possible syllable nuclei. However, this results in many peaks within each syllable. Therefore, the script then discards peaks that are not followed and preceded by dips of at least 2 dB in intensity. To discard loud fricatives as syllable nuclei, the script finally also discards all peaks that are not voiced. In the sound file attached is the actual speech file corresponding to the figure on this page.