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

Recently, the cascaded two-pass architecture has emerged as a strongcontender for on-device automatic speech recognition (ASR). A cascade of causaland shallow non-causal encoders coupled with a shared decoder enables operationin both streaming and look-ahead modes. In this paper, we propose shallowcascaded model by combining various model compression techniques such asknowledge distillation, shared decoder, and tied-and-reduced transducer networkin order to reduce the model footprint. The shared decoder is changed into atied-and-reduced network. The cascaded two-pass model is further compressedusing knowledge distillation using a Kullback-Leibler divergence loss on themodel posteriors. We demonstrate a 50% reduction in the size of a 41 Mparameter cascaded teacher model with no noticeable degradation in ASR accuracyand a 30% reduction in latency

On the compression of shallow non-causal ASR models using knowledge  distillation and tied-and-reduced decoder for low-latency on-device speech  recognition