editedbook
Publication: IIP Proceedings Year: 2022, Month: November Page No: 33-42, ISSN/ISBN: 978-93-95632-70-6, DOI/Link: https://rsquarel.org/assets/docupload/rsl2023788946E80941802.pdf
PHONOTACTIC AND PROSODIC FEATURES IN NEURAL NETWORK CLASSIFIERS FOR ASSAMESE DIALECT IDENTIFICATION
Area/Stream: Artificial Intelligenceaa, Authors: Hem Chandra Das Keywords: FFNN; dialect; phonotactic features; prosodic features; Assamese Book Name /series: Futuristic Trends in Artificial Intelligence, Volume 2, Book 16, Chapter 4Publication: IIP Proceedings Year: 2022, Month: November Page No: 33-42, ISSN/ISBN: 978-93-95632-70-6, DOI/Link: https://rsquarel.org/assets/docupload/rsl2023788946E80941802.pdf
Abstract:
In this study, we investigate phonotactic and prosodic features for dialect identification derived from the voice signal and its transcription. Feedforward Neural Network (FFNN) classifiers are trained to distinguish between dialects using the phonotactic and prosodic features at the trisyllabic level as a representation of dialect characteristics. We show that these traits do indeed include information specific to dialects. We also demonstrate the sufficiency of phonotactic features in terms of broad phonetic categories to express the phonotactic regularities/constraints dialects. For four Assamese dialecta, the effectiveness of the FFNN classifier based on these features is assessed.
Cite this: Hem Chandra Das,"PHONOTACTIC AND PROSODIC FEATURES IN NEURAL NETWORK CLASSIFIERS FOR ASSAMESE DIALECT IDENTIFICATION", Futuristic Trends in Artificial Intelligence, Volume 2, Book 16, Chapter 4, November, 2022, 33-42, 978-93-95632-70-6, https://rsquarel.org/assets/docupload/rsl2023788946E80941802.pdf
Views: 4168
Download File
News
