PATATRA and PATAFreq: two French databases for the documentation of within-speaker variability in speech

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 1939–1944 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 1939 PATATRA and PATAFreq: two French databases for the documentation of within-speaker variability in speech Cécile Fougeron*, Nicolas Audibert*, Cédric Gendrot*, Estelle Chardenon*1, Louise Wohmann* *Laboratoire de Phonétique et Phonologie (UMR7018 CNRS/Univ. Sorbonne-Nouvelle), 1Laboratoire Parole et Langage (UMR 7309, CNRS/AMU Univ.) *19 rue des Bernardins, 75005 Paris, France. {cecile.fougeron, Nicolas.audibert, cedric.gendrot, louise.wohmann}@sorbonne-nouvelle.fr estelle.chardenon@univ-amu.fr Abstract Our knowledge on speech is historically built on data comparing different speakers or data averaged across speakers. Consequently, little is known on the variability in the speech of a single individual. Experimental studies have shown that speakers adapt to the linguistic and the speaking contexts, and modify their speech according to their emotional or biological condition, etc. However, it is unclear how much speakers vary from one repetition to the next, and how comparable are recordings that are collected days, months or years apart. In this paper, we introduce two French databases which contain recordings of 9 to 11 speakers recorded over 9 to 18 sessions, allowing comparisons of speech tasks with a different delay between the repetitions: 3 repetitions within the same session, 6 to 10 repetitions on different days during a two months period, 5 to 9 repetitions on different years. Speakers are recorded on a large set of speech tasks including read and spontaneous speech as well as speech-like performance tasks. In this paper, we provide detailed descriptions of the two databases and available annotations. We conclude by an illustration on how these data can inform on within-speaker variability of speech. Keywords: inter- and intra-speaker variation, longitudinal, speech, French 1. Introduction Variability in speech is pervasive but it is also, at least in part, structured and ruled-governed. Therefore, documenting variation in speech and explaining its origins has been a longstanding endeavor in phonetic research. Two kinds of variability can be distinguished in terms of their origins and their characteristics: inter-talker and intra- talker variability (Wright 2006). On one side, inter-talker variability is rooted in speaker-specific physiological or anatomical traits, as well as demographical, regional or social factors (Foulkes & Docherty 2006). This inter-talker variability carries indexical information on the speaker: it is disseminated in an array of speech features and it is usually a bundle of speech features which are able to discriminate a speaker (or group of) from another. Furthermore, inter-talker variation is supposed to remain constant across speaking context. On the other side, intra- or within-taker variability originates from various sources (see for instance Bürki 2018 for a review). Among those, changes in the phonetic realization of phonemes, due to phonetic or prosodic contexts, has been the most documented and this within- talker variability is easy to model. This is not the case of other intra-talker sources of variability which are due to the fact that talkers adapt their speech to the speaking context, to its formality, to the social relation they have with …

Content truncated to 500 words