Revealing the Next Word and Character in Arabic: An Effective Blend of Long Short-Term Memory Networks and ARABERT

Arabic raw audio datasets were initially gathered to produce a corresponding signal spectrum, which was further used to extract the Mel-Frequency Cepstral Coefficients (MFCCs). The pronunciation dictionary, language model, and acoustic model were further derived from the MFCCs’ features. These outpu...

Full description

Saved in:
Bibliographic Details
Main Authors: Fawaz S. Al-Anzi, S. T. Bibin Shalini
Format: Article
Language:English
Published: MDPI AG 2024-11-01
Series:Applied Sciences
Subjects:
Online Access:https://www.mdpi.com/2076-3417/14/22/10498
Tags: Add Tag
No Tags, Be the first to tag this record!