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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

In a conventional Speech emotion recognition (SER) task, a classifier for agiven language is trained on a pre-existing dataset for that same language.However, where training data for a language does not exist, data from otherlanguages can be used instead. We experiment with cross-lingual andmultilingual SER, working with Amharic, English, German and URDU. For Amharic,we use our own publicly-available Amharic Speech Emotion Dataset (ASED). ForEnglish, German and Urdu we use the existing RAVDESS, EMO-DB and URDU datasets.We followed previous research in mapping labels for all datasets to just twoclasses, positive and negative. Thus we can compare performance on differentlanguages directly, and combine languages for training and testing. InExperiment 1, monolingual SER trials were carried out using three classifiers,AlexNet, VGGE (a proposed variant of VGG), and ResNet50. Results averaged forthe three models were very similar for ASED and RAVDESS, suggesting thatAmharic and English SER are equally difficult. Similarly, German SER is moredifficult, and Urdu SER is easier. In Experiment 2, we trained on one languageand tested on another, in both directions for each pair: Amharic<->German,Amharic<->English, and Amharic<->Urdu. Results with Amharic as target suggestedthat using English or German as source will give the best result. In Experiment3, we trained on several non-Amharic languages and then tested on Amharic. Thebest accuracy obtained was several percent greater than the best accuracy inExperiment 2, suggesting that a better result can be obtained when using two orthree non-Amharic languages for training than when using just one non-Amhariclanguage. Overall, the results suggest that cross-lingual and multilingualtraining can be an effective strategy for training a SER classifier whenresources for a language are scarce.

Cross-Corpus Multilingual Speech Emotion Recognition: Amharic vs. Other Languages