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

Machine lipreading (MLR) is speech recognition from visual cues and a nicheresearch problem in speech processing & computer vision. Current challengesfall into two groups: the content of the video, such as rate of speech or; theparameters of the video recording e.g, video resolution. We show that HD videois not needed to successfully lipread with a computer. The term "viseme" isused in machine lipreading to represent a visual cue or gesture whichcorresponds to a subgroup of phonemes where the phonemes are visuallyindistinguishable. A phoneme is the smallest sound one can utter, because thereare more phonemes per viseme, maps between units show a many-to-onerelationship. Many maps have been presented, we compare these and our resultsshow Lee's is best. We propose a new method of speaker-dependentphoneme-to-viseme maps and compare these to Lee's. Our results show thesensitivity of phoneme clustering and we use our new knowledge to augment aconventional MLR system. It has been observed in MLR, that classifiers needtraining on test subjects to achieve accuracy. Thus machine lipreading ishighly speaker-dependent. Conversely speaker independence is robustclassification of non-training speakers. We investigate the dependence ofphoneme-to-viseme maps between speakers and show there is not a highvariability of visemes, but there is high variability in trajectory betweenvisemes of individual speakers with the same ground truth. This implies adependency upon the number of visemes within each set for each individual. Weshow that prior phoneme-to-viseme maps rarely have enough visemes and theoptimal size, which varies by speaker, ranges from 11-35. Finally we decodefrom visemes back to phonemes and into words. Our novel approach uses theoptimum range visemes within hierarchical training of phoneme classifiers anddemonstrates a significant increase in classification accuracy.

Decoding visemes: improving machine lipreading