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

Modern Automatic Speech Recognition (ASR) systems can achieve highperformance in terms of recognition accuracy. However, a perfectly accuratetranscript still can be challenging to read due to disfluency, filter words,and other errata common in spoken communication. Many downstream tasks andhuman readers rely on the output of the ASR system; therefore, errorsintroduced by the speaker and ASR system alike will be propagated to the nexttask in the pipeline. In this work, we propose an ASR post-processing modelthat aims to transform the incorrect and noisy ASR output into a readable textfor humans and downstream tasks. We leverage the Metadata Extraction (MDE)corpus to construct a task-specific dataset for our study. Since the dataset issmall, we propose a novel data augmentation method and use a two-stage trainingstrategy to fine-tune the RoBERTa pre-trained model. On the constructed testset, our model outperforms a production two-step pipeline-based post-processingmethod by a large margin of 13.26 on readability-aware WER (RA-WER) and 17.53on BLEU metrics. Human evaluation also demonstrates that our method cangenerate more human-readable transcripts than the baseline method.

Generating Human Readable Transcript for Automatic Speech Recognition with Pre-trained Language Model