Can I read End-to-End Arabic Dialect Recognition on EtoBox?

End-to-End Arabic Dialect Recognition by shivang dhiman is a document available to read on EtoBox.

What is End-to-End Arabic Dialect Recognition about?

This paper presents an end-to-end dialect identification (DID) system utilizing convolutional neural networks and a Siamese network for language embeddings, specifically targeting Arabic dialects. The system employs acoustic features like MFCCs and FBANK, along with linguistic features, achieving a maximum accuracy of 78% on the MGB-3 dataset. The study highlights the challenges of DID due to the linguistic similarities among dialects and explores dataset augmentation techniques to enhance performance.

Author

shivang dhiman

Language

EN