Dataset Information

Correlated lip motion and voice audio data.

ABSTRACT: This data set is comprised of correlated audio and lip movement data in multiple videos of multiple subjects reading the same text. It was collected to facilitate the development and validation of algorithms used to train and test a compound biometric system that consists of lip-motion and voice recognition. The data set is a collection of videos of volunteers reciting a fixed script that is intended to be used to train software to recognize voice and lip-motion patterns. A second video is included of the individual reciting a shorter phrase, which is designed to be used to test the recognition functionality of the system. The recordings were collected in a controlled, indoor setting with a 4K professional-grade camcorder and adjustable, LED lights.

SUBMITTER: Colasito M

PROVIDER: S-EPMC6218630 | biostudies-literature | 2018 Dec

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Correlated lip motion and voice audio data.

Colasito Marco M Straub Jeremy J Kotala Pratap P

Data in brief 20181018

This data set is comprised of correlated audio and lip movement data in multiple videos of multiple subjects reading the same text. It was collected to facilitate the development and validation of algorithms used to train and test a compound biometric system that consists of lip-motion and voice recognition. The data set is a collection of videos of volunteers reciting a fixed script that is intended to be used to train software to recognize voice and lip-motion patterns. A second video is inclu ...[more]

PMID: 30417045

Similar Datasets

Project description:Motor-task functional magnetic resonance imaging (fMRI) is crucial in the study of several clinical conditions, including stroke and Parkinson's disease. However, motor-task fMRI is complicated by task-correlated head motion, which can be magnified in clinical populations and confounds motor activation results. One method that may mitigate this issue is multi-echo independent component analysis (ME-ICA), which has been shown to separate the effects of head motion from the desired BOLD signal but has not been tested in motor-task datasets with high amounts of motion. In this study, we collected an fMRI dataset from a healthy population who performed a hand grasp task with and without task-correlated amplified head motion to simulate a motor-impaired population. We analyzed these data using three models: single-echo (SE), multi-echo optimally combined (ME-OC), and ME-ICA. We compared the models' performance in mitigating the effects of head motion on the subject level and group level. On the subject level, ME-ICA better dissociated the effects of head motion from the BOLD signal and reduced noise. Both ME models led to increased t-statistics in brain motor regions. In scans with high levels of motion, ME-ICA additionally mitigated artifacts and increased stability of beta coefficient estimates, compared to SE. On the group level, all three models produced activation clusters in expected motor areas in scans with both low and high motion, indicating that group-level averaging may also sufficiently resolve motion artifacts that vary by subject. These findings demonstrate that ME-ICA is a useful tool for subject-level analysis of motor-task data with high levels of task-correlated head motion. The improvements afforded by ME-ICA are critical to improve reliability of subject-level activation maps for clinical populations in which group-level analysis may not be feasible or appropriate, for example in a chronic stroke cohort with varying stroke location and degree of tissue damage.

Dataset Information

Correlated lip motion and voice audio data.

Publications

Correlated lip motion and voice audio data.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets