Latest Advances in Computational Speech Analysis for Mobile Sensing

Cummins, Nicholas; Schuller, Björn W.

doi:10.1007/978-3-030-31620-4_9

Cited by 3 publications

(2 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…OpenSmile (Eyben et al, 2010) is the de-facto state of the art for feature extraction (Cummins & Schuller, 2019). Specifically, it is still more popular than learned representation in the areas of emotion recognition (Latif et al, 2020) and many medical tasks.…”

Section: Opensmile Featuresmentioning

confidence: 99%

Towards Learning a Universal Non-Semantic Representation of Speech

Shor,

Jansen,

Maor

et al. 2020

Preprint

View full text Add to dashboard Cite

The ultimate goal of transfer learning is to reduce labeled data requirements by exploiting a preexisting embedding model trained for different datasets or tasks. While significant progress has been made in the visual and language domains, the speech community has yet to identify a strategy with wide-reaching applicability across tasks. This paper describes a representation of speech based on an unsupervised triplet-loss objective, which exceeds state-of-the-art performance on a number of transfer learning tasks drawn from the non-semantic speech domain. The embedding is trained on a publicly available dataset, and it is tested on a variety of low-resource downstream tasks, including personalization tasks and medical domain. The model will be publicly released.

show abstract

Section: Opensmile Featuresmentioning

confidence: 99%

Towards Learning a Universal Non-Semantic Representation of Speech

Shor,

Jansen,

Maor

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…Auditory nonverbal disclosures are noisy signals, and thus inferences inherently fall short of precision, both when made by humans (Juslin and Scherer 2005) and when made by algorithms (Kröger, Lutz and Raschke 2020). Yet, auditory signal analysis is a highly active field of research, in which both the scope of investigated phenomena and performance accuracy keep increasing 5 (for technical reviews of the state of the art, see, e.g., Bai et al [2020], Chaki [2021], Cummins and Schuller [2019], Schuller et al [2021], and Singh [2019]). All computational approaches referenced in Table 2 perform above chance level; in certain domains, algorithmic inferences based on auditory nonverbal disclosures are already as accurate as, and sometimes even more accurate than, human judgments (Cummins, Baird, and Schuller 2018; Kröger, Lutz, and Raschke 2020).…”

Section: Nonverbal Disclosure In Oral Versus Manual Interactions With...mentioning

confidence: 99%

Information Disclosure in the Era of Voice Technology

Melzner¹,

Bonezzi²,

Meyvis³

2023

Journal of Marketing

View full text Add to dashboard Cite

Whenever consumers interact with technological devices connected to the internet, they disclose information about themselves. The rapid diffusion of voice technology is shifting the way consumers interact with technological devices from typing or clicking to speaking. This article offers a comprehensive analysis of how this occurring shift from manual to oral communication with technology affects information disclosure. It first considers verbal disclosure and provides a conceptual framework that explicates how voice technology can influence consumers’ propensity to reveal information about themselves through semantic content disclosed voluntarily. It then considers non-verbal disclosure and provides an analysis of how voice technology enables the collection of information revealed rather unintentionally through vocal paralanguage and ambient sound. The article offers testable propositions and poses open research questions that can serve as impetus for future research. Additionally, it provides insights to marketers for how to navigate voice technology as a source of consumer information and to policymakers for how to better protect consumer privacy in interactions with voice technology.

show abstract