AphasiaBD: A Bengali Speech Dataset for Broca's and Global Aphasia

AphasiaBD is a Bengali speech dataset collected from 38 post-stroke aphasia patients diagnosed with Broca's Aphasia or Global Aphasia. The dataset contains 335 segmented audio clips recorded during multiple speech tasks, including conversation, repetition, question answering, and reading. Participants were recruited from diverse geographical regions of Bangladesh to capture variations in accent and speech characteristics. The recordings were collected using smartphone devices in natural environments and manually segmented at the sentence level. The dataset reflects a wide range of aphasia-induced speech impairments, including phonemic distortions, word substitutions, prolonged pauses, and incomplete utterances, making it a valuable resource for research on automatic speech recognition, speech correction, aphasia detection, and low-resource Bengali speech processing.AphasiaBD is a Bengali speech dataset collected from 38 post-stroke aphasia patients diagnosed with Broca's Aphasia or Global Aphasia. The dataset contains 335 segmented audio clips recorded during multiple speech tasks, including conversation, repetition, question answering, and reading. Participants were recruited from diverse geographical regions of Bangladesh to capture variations in accent and speech characteristics. The recordings were collected using smartphone devices in natural environments and manually segmented at the sentence level. The dataset reflects a wide range of aphasia-induced speech impairments, including phonemic distortions, word substitutions, prolonged pauses, and incomplete utterances, making it a valuable resource for research on automatic speech recognition, speech correction, aphasia detection, and low-resource Bengali speech processing.