In this paper, we investigate several existing and a new stateof-the-art generative adversarial network-based (GAN) voice conversion method for enhancing dysarthric speech for improved dysarthric speech recognition. We compare key components of existing methods as part of a rigorous ablation study to find the most effective solution to improve dysarthric speech recognition. We find that straightforward signal processing methods such as stationary noise removal and vocoder-based time stretching lead to dysarthric speech recognition results comparable to those obtained when using state-of-the-art GANbased voice conversion methods as measured using a phoneme recognition task. Additionally, our proposed solution of a combination of MaskCycleGAN-VC and time stretched enhancement is able to improve the phoneme recognition results for certain dysarthric speakers compared to our time stretched baseline.

This study aims to interpret the early-stage rheology of alkali-activated slag (AAS) paste from microstructure perspectives. The microstructures visualized by cryogenic scanning electron microscopy (cryo-SEM) revealed the essential distinction between hydroxide and silicate-activated slag pastes. The hydroxide-based mixture showed typical suspension features, where slag particles were dispersed in the hydroxide activators. In the hydroxide media, even at very early ages (5 min), the solid grains were attached to each other through rigid connections of reaction products, which resulted in high yield stress. As for the silicate-based mixtures, an emulsion phase has been observed between slag particles, which consists of discontinuous water droplets and continuous silicate gels. Fine emulsions with smaller water droplets were observed as the silicate modulus of activators increased, which dispersed the slag particles but on the other hand improved the viscosity of the paste. With increasing water to binder ratio, both yield stress and viscosity of AAS pastes significantly reduced.