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

In this paper, we propose a language-universal adapter learning frameworkbased on a pre-trained model for end-to-end multilingual automatic speechrecognition (ASR). For acoustic modeling, the wav2vec 2.0 pre-trained model isfine-tuned by inserting language-specific and language-universal adapters. Anonline knowledge distillation is then used to enable the language-universaladapters to learn both language-specific and universal features. The linguisticinformation confusion is also reduced by leveraging language identifiers(LIDs). With LIDs we perform a position-wise modification on the multi-headattention outputs. In the inference procedure, the language-specific adaptersare removed while the language-universal adapters are kept activated. Theproposed method improves the recognition accuracy and addresses the linearincrease of the number of adapters' parameters with the number of languages incommon multilingual ASR systems. Experiments on the BABEL dataset confirm theeffectiveness of the proposed framework. Compared to the conventionalmultilingual model, a 3.3% absolute error rate reduction is achieved. The codeis available at: https://github.com/shen9712/UniversalAdapterLearning.

Language-Universal Adapter Learning with Knowledge Distillation for End-to-End Multilingual Speech Recognition