The Automatic Search for Sounding Segments of SPPAS: Application to Cheese! Corpus

Abstract

The development of corpora inevitably involves the need for segmentation. For most of the corpora, the first segmentation to operate consist in determining silences vs Inter-Pausal Units -IPUs, i.e. sounding segments. This paper presents the "Search for IPUs" feature included in SPPAS -the automatic annotation and analysis of speech software tool distributed under the terms of public licenses. Particularly, this paper is focusing on its evaluation on Cheese! corpus, a corpus of reading then conversational speech between two participants. The paper reports the number of manual actions which was performed manually by the annotators in order to obtain the expected segmentation: add new IPUs, ignore irrelevant ones, split an IPU, merge two consecutive ones and move boundaries. The evaluation shows that the proposed fully automatic method is relevant.

Speech markers of Cancer-Related Cognitive Impairment: A pilot study - Speech Prosody 2024

“…All samples were automatically transcribed with Whisper [15] and manually checked by the first author. Pause annotation was performed with the SPPAS software [16] and errors were manually corrected by the first author in Praat. In total, we annotated 874 pauses for the picture task and 1355 pauses for the memory task.…”

MUST&P-SRL: Multi-lingual and Unified Syllabification in Text and Phonetic Domains for Speech Representation Learning

“…The linguistic literature identified exceptions to this principle, the main one being probably the sibilant-stop consonant cluster (Iacoponi and Savy, 2011;Yin et al, 2023;DeLisi, 2015). Implementations of the principle with processing of these exceptions been successfully applied for automatic syllabification in several languages in the pronunciation domain with very high word accuracies (Bigi et al, 2010;Bigi and Petrone, 2014;Bigi and Klessa, 2015). But it has also been applied in the spelling domain with some success for some languages.…”

Transcribing and Aligning Conversational Speech: A Hybrid Pipeline Applied to French Conversations

“…The SPPAS IPU-detection algorithm [15] allows for an optimized threshold value for each file. First, it calculates the root-mean-square (RMS) of the intensity inside a sliding time window of duration 20 ms and then calculates the threshold value Θ = min +µ − 1.5σ where min is the minimum, µ is the mean, and σ is the standard deviation for RMS values.…”