A multispeaker dataset of raw and reconstructed speech production real-time MRI video and 3D volumetric images

本数据集由南加州大学创建，包含75名参与者的原始和重建的实时MRI视频及3D体积图像，用于研究语音产生。数据集涵盖了多种语言学激励的语音任务，并首次公开了优化的语音产生实验设置下的原始多线圈实时MRI数据。此外，数据集还包括在持续语音声音期间的3D体积声道MRI和高分辨率静态解剖T2加权上呼吸道MRI。该数据集的应用领域广泛，包括语音科学、语言学、生物启发语音技术开发和临床应用，旨在解决语音产生的动态图像重建、伪影校正、特征提取和语言学相关生物标志物的直接提取等问题。

This dataset was created by the University of Southern California. It encompasses raw and reconstructed real-time MRI videos and 3D volumetric images from 75 participants, for use in speech production research. The dataset includes a diverse set of linguistically motivated speech tasks, and marks the first public release of raw multi-coil real-time MRI data acquired under optimized experimental settings for speech production studies. Additionally, the dataset features 3D volumetric vocal tract MRI captured during sustained phonation, alongside high-resolution static anatomical T2-weighted upper airway MRI. This resource has broad applications across speech science, linguistics, bio-inspired speech technology development, and clinical settings, and aims to address core challenges including dynamic image reconstruction for speech production, artifact correction, feature extraction, and direct extraction of linguistically relevant biomarkers.