A Survey on perceived speaker traits: Personality, likability, pathology, and the first challenge

Abstract

The INTERSPEECH 2012 Speaker Trait Challenge aimed at a unified test-bed for perceived speaker traits -the first challenge of this kind: personality in the five OCEAN personality dimensions, likability of speakers, and intelligibility of pathologic speakers. In the present article, we give a brief overview of the state-of-the-art in these three fields of research and describe the three sub-challenges in terms of the challenge conditions, the baseline results provided by the organisers, and a new openSMILE feature set, which has been used for computing the baselines and which has been provided to the participants. Furthermore, we summarise the approaches and the results presented by the participants to show the various techniques that are currently applied to solve these classification tasks.

Acoustic-Prosodic Automatic Personality Trait Assessment for Adults and Children

“…The SPC corpus was used in the Interspeech 2012 Speaker Trait ChallengePersonality Sub-challenge [19,20]. On that occasion, the corpus was split into 3 speaker-independent train, development and test subsets consisting of 256, 183 and 201 files, respectively.…”

“…In this work, the same experimental setup as that employed in the Interspeech 2012 Speaker Trait Challenge-Personality Sub-challenge [19,20] has been adopted. The models used in this work are linear support vector machines (SVM) trained by means of the well-known SMO algorithm.…”

“…Two very different speech corpora have been used in this work, namely the Speaker Personality Corpus (SPC) [13,19,20] and the Game-of-Nines (GoN) corpus [2]. The markedly different characteristics of the speakers in these two corpora (adults vs. children, French vs. Portuguese speakers, respectively) constitute a good basis to investigate whether we can leverage the presence of common, language-and age-independent acoustic and prosodic cues to detect/assess ba-sic personality traits across heterogeneous groups of target speakers.…”

State-of-the-art tasks and achievements of paralinguistic speech analysis systems

“…В 2012 году соревнования Speaker Trait Challenge проходили по направлениям (базовая статья [26]): распознавание индивидуальности диктора (Personality Sub-Challenge) [27], оценка привлекательно-сти голоса диктора (Likability Sub-Challenge) [28] и определение патологии речи (Pathology SubChallenge) [29]. В табл.…”

“…1 представлена информация о результатах и работах победителей и призеров дан-ных трех конкурсов, а также сравнение с базовыми результатами организаторов. В данном соревновании все результаты (качество работы автоматических систем классификации акустических паралингвистиче-ских явлений) оцениваются и сравниваются с применением основной метрики полноты (Unweighted Average Recall, UAR) [26]. [36], определение наличия конфликта между дикторами (Conflict Sub-Challenge) [37], распознавание эмоционального состояния диктора (Emotion Sub-Challenge) [38], анализ социальных сигналов (Social Signals Sub-Challenge) [39].…”