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

Most state-of-the-art speech systems are using Deep Neural Networks (DNNs).Those systems require a large amount of data to be learned. Hence, learningstate-of-the-art frameworks on under-resourced speech languages/problems is adifficult task. Problems could be the limited amount of data for impairedspeech. Furthermore, acquiring more data and/or expertise is time-consuming andexpensive. In this paper we position ourselves for the following speechprocessing tasks: Automatic Speech Recognition, speaker identification andemotion recognition. To assess the problem of limited data, we firstlyinvestigate state-of-the-art Automatic Speech Recognition systems as itrepresents the hardest tasks (due to the large variability in each language).Next, we provide an overview of techniques and tasks requiring fewer data. Inthe last section we investigate few-shot techniques as we interpretunder-resourced speech as a few-shot problem. In that sense we propose anoverview of few-shot techniques and perspectives of using such techniques forthe focused speech problems in this survey. It occurs that the reviewedtechniques are not well adapted for large datasets. Nevertheless, somepromising results from the literature encourage the usage of such techniquesfor speech processing.

Deep Neural Networks for Automatic Speech Processing: A Survey from Large Corpora to Limited Data