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

Acoustics-to-word models are end-to-end speech recognizers that use words astargets without relying on pronunciation dictionaries or graphemes. Thesemodels are notoriously difficult to train due to the lack of linguisticknowledge. It is also unclear how the amount of training data impacts theoptimization and generalization of such models. In this work, we study theoptimization and generalization of acoustics-to-word models under differentamounts of training data. In addition, we study three types of inductive bias,leveraging a pronunciation dictionary, word boundary annotations, andconstraints on word durations. We find that constraining word durations leadsto the most improvement. Finally, we analyze the word embedding space learnedby the model, and find that the space has a structure dominated by thepronunciation of words. This suggests that the contexts of words, instead oftheir phonetic structure, should be the future focus of inductive bias inacoustics-to-word models.

On The Inductive Bias of Words in Acoustics-to-Word Models