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

Despite the rapid progress of automatic speech recognition (ASR) technologiestargeting normal speech in recent decades, accurate recognition of dysarthricand elderly speech remains highly challenging tasks to date. Sources ofheterogeneity commonly found in normal speech including accent or gender, whenfurther compounded with the variability over age and speech pathology severitylevel, create large diversity among speakers. To this end, speaker adaptationtechniques play a key role in personalization of ASR systems for such users.Motivated by the spectro-temporal level differences between dysarthric, elderlyand normal speech that systematically manifest in articulatory imprecision,decreased volume and clarity, slower speaking rates and increased dysfluencies,novel spectrotemporal subspace basis deep embedding features derived using SVDspeech spectrum decomposition are proposed in this paper to facilitateauxiliary feature based speaker adaptation of state-of-the-art hybrid DNN/TDNNand end-to-end Conformer speech recognition systems. Experiments were conductedon four tasks: the English UASpeech and TORGO dysarthric speech corpora; theEnglish DementiaBank Pitt and Cantonese JCCOCC MoCA elderly speech datasets.The proposed spectro-temporal deep feature adapted systems outperformedbaseline i-Vector and xVector adaptation by up to 2.63% absolute (8.63%relative) reduction in word error rate (WER). Consistent performanceimprovements were retained after model based speaker adaptation using learninghidden unit contributions (LHUC) was further applied. The best speaker adaptedsystem using the proposed spectral basis embedding features produced the lowestpublished WER of 25.05% on the UASpeech test set of 16 dysarthric speakers.

Speaker Adaptation Using Spectro-Temporal Deep Features for Dysarthric and Elderly Speech Recognition