Speech-Based Bangladeshi Dialect Detection for Accurate Transcription Systems

Published: 5 August 2026| Version 1 | DOI: 10.17632/2ttsgvcwws.1
Contributors:
Md Mijanur Rahman,
,
,
,

Description

The dataset consists of short speech segments collected from two main sources: 1. Online Sources - Audio extracted from publicly available videos on YouTube - Speakers from different regions - Varied background noise and speaking styles 2. Real-Life Recordings - Recorded from native speakers - Controlled environment - Clear pronunciation of dialect-specific speech 3. Audio Processing - Stored in structured folders by dialect - The total number of samples = almost 4000 audio clips - Included dialects: Standard Bengali, Sylheti, Chittagong, Noakhali, Mymensingh - Format: WAV - Duration: each clip is about 4 to 12 seconds long

Files

Institutions

Categories

Computer Science, Artificial Intelligence, Speech Processing, Computational Linguistics, Data Science, Natural Language Processing, Speech Recognition, Machine Learning, Deep Learning

Licence