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

Pre-trained speech representations like wav2vec 2.0 are a powerful tool forautomatic speech recognition (ASR). Yet many endangered languages lacksufficient data for pre-training such models, or are predominantly oralvernaculars without a standardised writing system, precluding fine-tuning.Query-by-example spoken term detection (QbE-STD) offers an alternative foriteratively indexing untranscribed speech corpora by locating spoken queryterms. Using data from 7 Australian Aboriginal languages and a regional varietyof Dutch, all of which are endangered or vulnerable, we show that QbE-STD canbe improved by leveraging representations developed for ASR (wav2vec 2.0: theEnglish monolingual model and XLSR53 multilingual model). Surprisingly, theEnglish model outperformed the multilingual model on 4 Australian languagedatasets, raising questions around how to optimally leverage self-supervisedspeech representations for QbE-STD. Nevertheless, we find that wav2vec 2.0representations (either English or XLSR53) offer large improvements (56-86%relative) over state-of-the-art approaches on our endangered language datasets.

Leveraging pre-trained representations to improve access to untranscribed speech from endangered languages