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 modern interactive speech-based systems, speech is consumed andtranscribed incrementally prior to having disfluencies removed. Thispost-processing step is crucial for producing clean transcripts and highperformance on downstream tasks (e.g. machine translation). However, mostcurrent state-of-the-art NLP models such as the Transformer operatenon-incrementally, potentially causing unacceptable delays. We propose astreaming BERT-based sequence tagging model that, combined with a noveltraining objective, is capable of detecting disfluencies in real-time whilebalancing accuracy and latency. This is accomplished by training the model todecide whether to immediately output a prediction for the current input or towait for further context. Essentially, the model learns to dynamically size itslookahead window. Our results demonstrate that our model produces comparablyaccurate predictions and does so sooner than our baselines, with lower flicker.Furthermore, the model attains state-of-the-art latency and stability scoreswhen compared with recent work on incremental disfluency detection.

Teaching BERT to Wait: Balancing Accuracy and Latency for Streaming Disfluency Detection