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

Streaming end-to-end speech recognition models have been widely applied tomobile devices and show significant improvement in efficiency. These models aretypically trained on the server using transcribed speech data. However, theserver data distribution can be very different from the data distribution onuser devices, which could affect the model performance. There are two mainchallenges for on device training, limited reliable labels and limited trainingmemory. While self-supervised learning algorithms can mitigate the mismatchbetween domains using unlabeled data, they are not applicable on mobile devicesdirectly because of the memory constraint. In this paper, we propose anincremental layer-wise self-supervised learning algorithm for efficient speechdomain adaptation on mobile devices, in which only one layer is updated at atime. Extensive experimental results demonstrate that the proposed algorithmobtains a Word Error Rate (WER) on the target domain $24.2\%$ better thansupervised baseline and costs $89.7\%$ less training memory than the end-to-endself-supervised learning algorithm.

Incremental Layer-wise Self-Supervised Learning for Efficient Speech Domain Adaptation On Device