English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Modelling of early language acquisition aims to understand how infantsbootstrap their language skills. The modelling encompasses properties of theinput data used for training the models, the cognitive hypotheses and theiralgorithmic implementations being tested, and the evaluation methodologies tocompare models to human data. Recent developments have enabled the use of morenaturalistic training data for computational models. This also motivatesdevelopment of more naturalistic tests of model behaviour. A crucial steptowards such an aim is to develop representative speech datasets consisting ofspeech heard by infants in their natural environments. However, a majordrawback of such recordings is that they are typically noisy, and it iscurrently unclear how the sound quality could affect analyses and modellingexperiments conducted on such data. In this paper, we explore this aspect forthe case of infant-directed speech (IDS) and adult-directed speech (ADS)analysis. First, we manually and automatically annotated audio quality ofutterances extracted from two corpora of child-centred long-form recordings (inEnglish and French). We then compared acoustic features of IDS and ADS in anin-lab dataset and across different audio quality subsets of naturalistic data.Finally, we assessed how the audio quality and recording environment may changethe conclusions of a modelling analysis using a recent self-supervised learningmodel. Our results show that the use of modest and high audio qualitynaturalistic speech data result in largely similar conclusions on IDS and ADSin terms of acoustic analyses and modelling experiments. We also found that anautomatic sound quality assessment tool can be used to screen out useful partsof long-form recordings for a closer analysis with comparable results to thatof manual quality annotation.

Analysing the Impact of Audio Quality on the Use of Naturalistic Long-Form Recordings for Infant-Directed Speech Research