Praat script to detect syllable nuclei and measure speech rate automatically

Area of Science:

• Phonetics and Speech Science
• Computational Linguistics
• Acoustic Phonetics

Background:

• Accurate speech rate measurement is crucial for various linguistic and clinical applications.
• Manual transcription for syllable counting is time-consuming and labor-intensive.
• Existing automated methods may lack precision or require extensive linguistic expertise.

Purpose of the Study:

• To develop and validate an automated method for detecting syllable nuclei.
• To enable reliable speech rate measurement without manual transcription.
• To assess the efficacy of automated syllable counting in spoken Dutch corpora.

Main Methods:

• A script in Praat software was developed to identify syllable nuclei.
• Potential nuclei were identified as intensity peaks preceded and followed by dips.
• Unvoiced peaks were excluded to refine syllable nucleus detection.

Main Results:

• The automated script demonstrated high correlations with human-derived syllable counts.
• Speech rate calculated using automated syllable counts closely matched human-based calculations.
• The method proved effective on two corpora of spoken Dutch.

Conclusions:

• Automated syllable nucleus detection provides a reliable measure of speech rate.
• This method eliminates the need for manual transcription, saving time and resources.
• The validated approach is suitable for comparing speech rates across different participants and tasks.