Edited by Shlomo Izre'el, Heliana Mello, Alessandro Panunzi and Tommaso Raso
[Studies in Corpus Linguistics 94] 2020
► pp. 285–300
Chapter 9Cross-linguistic comparison of automatic detection of speech breaks in read and
narrated speech in four languages
This chapter tests an algorithm for the automatic detection of speech breaks in read and narrated speech in Brazilian Portuguese (BP), European Portuguese (EP), French, and German. The algorithm is independent of previous transcription or linguistic analysis (syllable, phone labeling and segmentation), requiring only the audio file. It operates in two stages: vowel onsets detection firstly, followed by V-to-V duration intervals normalization for smoothed duration z-scores. Peaks over 2.5 of the latter were considered speech breaks. Compared to human segmentation, hits for reading (70%) were higher than for narration (60%). Crosslinguistic results show EP and French having the highest proportion of hits. A test with the English Navy audio file reveals a hit proportion similar to German.
- 2.2The SalienceDetector script
- 3.1Testing with English spontaneous speech
Cited by 1 other publications
This list is based on CrossRef data as of 28 december 2021. Please note that it may not be complete. Sources presented here have been supplied by the respective publishers. Any errors therein should be reported to them.