ABSTRACT
Video represents the majority of internet traffic today, driving a continual race between the generation of higher quality content, transmission of larger file sizes, and the development of network infrastructure. In addition, the recent COVID-19 pandemic fueled a surge in the use of video conferencing tools. Since videos take up considerable bandwidth ( ∼ 100 Kbps to a few Mbps), improved video compression can have a substantial impact on network performance for live and pre-recorded content, providing broader access to multimedia content worldwide. We present a novel video compression pipeline, called Txt2Vid, which dramatically reduces data transmission rates by compressing webcam videos ('talking-head videos') to a text transcript. The text is transmitted and decoded into a realistic reconstruction of the original video using recent advances in deep learning based voice cloning and lip syncing models. Our generative pipeline achieves two to three orders of magnitude reduction in the bitrate as compared to the standard audio-video codecs (encoders-decoders), while maintaining equivalent Quality-of-Experience based on a subjective evaluation by users ( n=242 ) in an online study. The Txt2Vid framework opens up the potential for creating novel applications such as enabling audio-video communication during poor internet connectivity, or in remote terrains with limited bandwidth. The code for this work is available at https://github.com/tpulkit/txt2vid.git. © 1983-2012 IEEE.
ABSTRACT
Video represents the majority of internet traffic today, driving a continual race between the generation of higher quality content, transmission of larger file sizes, and the development of network infrastructure. In addition, the recent COVID-19 pandemic fueled a surge in the use of video conferencing tools. Since videos take up considerable bandwidth (~100 Kbps to a few Mbps), improved video compression can have a substantial impact on network performance for live and pre-recorded content, providing broader access to multimedia content worldwide. We present a novel video compression pipeline, called Txt2Vid, which dramatically reduces data transmission rates by compressing webcam videos (“talking-head videos”) to a text transcript. The text is transmitted and decoded into a realistic reconstruction of the original video using recent advances in deep learning based voice cloning and lip syncing models. Our generative pipeline achieves two to three orders of magnitude reduction in the bitrate as compared to the standard audio-video codecs (encoders-decoders), while maintaining equivalent Quality-of-Experience based on a subjective evaluation by users (n = 242) in an online study. The Txt2Vid framework opens up the potential for creating novel applications such as enabling audio-video communication during poor internet connectivity, or in remote terrains with limited bandwidth. The code for this work is available at https://github.com/tpulkit/txt2vid.git. IEEE
ABSTRACT
The home is a place of shelter, a place for family, and for separation from other parts of life, such as work. Global challenges, the most pressing of which are currently the COVID-19 pandemic and climate change has forced extra roles into many homes and will continue to do so in the future. Biodesign integrates living organisms into designed solutions and can offer opportunities for new kinds of technologies to facilitate a transition to the home of the future. Many families have had to learn to work alongside each other, and technology has mediated a transition from standard models of operation for industries. These are the challenges of the 21st century that mandate careful thinking around interactive systems and innovations that support new ways of living and working at home. In this workshop, we will explore opportunities for biodesign interactive systems in the future home. We will bring together a broad group of researchers in HCI, design, and biosciences to build the biodesign community and discuss speculative design futures. The outcome will generate an understanding of the role of interactive biodesign systems at home, as a place with extended functionalities.
ABSTRACT
Case fatality rate (CFR) and doubling time are important characteristics of any epidemic. For coronavirus disease 2019 (COVID-19), wide variations in the CFR and doubling time have been noted among various countries. Early in the epidemic, CFR calculations involving all patients as denominator do not account for the hospitalised patients who are ill and will die in the future. Hence, we calculated cumulative CFR (cCFR) using only patients whose final clinical outcomes were known at a certain time point. We also estimated the daily average doubling time. Calculating CFR using this method leads to temporal stability in the fatality rates, the cCFR stabilises at different values for different countries. The possible reasons for this are an improved outcome rate by the end of the epidemic and a wider testing strategy. The United States, France, Turkey and China had high cCFR at the start due to low outcome rate. By 22 April, Germany, China and South Korea had a low cCFR. China and South Korea controlled the epidemic and achieved high doubling times. The doubling time in Russia did not cross 10 days during the study period.