Bangla speech dataset

The Bangla speech dataset contains 3.85k hours of Bangla speech recordings with corresponding text labels and is used to evaluate the performance of text-to-speech systems in generating natural and intelligible synthesized Bangla speech.

Papers using Bangla speech dataset (2)

BnTTS: Few-Shot Speaker Adaptation in Low-Resource Setting2025 · 2 citesPseudo-Labeling for Domain-Agnostic Bangla Automatic Speech Recognition2023