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

Automatic recognition systems for child speech are lagging behind thosededicated to adult speech in the race of performance. This phenomenon is due tothe high acoustic and linguistic variability present in child speech caused bytheir body development, as well as the lack of available child speech data.Young readers speech additionally displays peculiarities, such as slow readingrate and presence of reading mistakes, that hardens the task. This workattempts to tackle the main challenges in phone acoustic modelling for youngchild speech with limited data, and improve understanding of strengths andweaknesses of a wide selection of model architectures in this domain. We findthat transfer learning techniques are highly efficient on end-to-endarchitectures for adult-to-child adaptation with a small amount of child speechdata. Through transfer learning, a Transformer model complemented with aConnectionist Temporal Classification (CTC) objective function, reaches a phoneerror rate of 28.1%, outperforming a state-of-the-art DNN-HMM model by 6.6%relative, as well as other end-to-end architectures by more than 8.5% relative.An analysis of the models' performance on two specific reading tasks (isolatedwords and sentences) is provided, showing the influence of the utterance lengthon attention-based and CTC-based models. The Transformer+CTC model displays anability to better detect reading mistakes made by children, that can beattributed to the CTC objective function effectively constraining the attentionmechanisms to be monotonic.

End-to-end acoustic modelling for phone recognition of young readers