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

Current state-of-the-art automatic speech recognition systems are trained towork in specific `domains', defined based on factors like application, samplingrate and codec. When such recognizers are used in conditions that do not matchthe training domain, performance significantly drops. This work explores theidea of building a single domain-invariant model for varied use-cases bycombining large scale training data from multiple application domains. Ourfinal system is trained using 162,000 hours of speech. Additionally, eachutterance is artificially distorted during training to simulate effects likebackground noise, codec distortion, and sampling rates. Our results show that,even at such a scale, a model thus trained works almost as well as thosefine-tuned to specific subsets: A single model can be robust to multipleapplication domains, and variations like codecs and noise. More importantly,such models generalize better to unseen conditions and allow for rapidadaptation -- we show that by using as little as 10 hours of data from a newdomain, an adapted domain-invariant model can match performance of adomain-specific model trained from scratch using 70 times as much data. We alsohighlight some of the limitations of such models and areas that need addressingin future work.

Toward domain-invariant speech recognition via large scale training