Community-Supported Shared Infrastructure in Support of Speech Accessibility.

Purpose: The Speech Accessibility Project (SAP) intends to facilitate research and development in automatic speech recognition (ASR) and other machine learning tasks for people with speech disabilities. The purpose of this article is to introduce this project as a resource for researchers, including baseline analysis of the first released data package.

Method: The project aims to facilitate ASR research by collecting, curating, and distributing transcribed U.

Fast and optimal multiframe blind deconvolution algorithm for high-resolution ground-based imaging of space objects.

We report a multiframe blind deconvolution algorithm that we have developed for imaging through the atmosphere. The algorithm has been parallelized to a significant degree for execution on high-performance computers, with an emphasis on distributed-memory systems so that it can be hosted on commodity clusters. As a result, image restorations can be obtained in seconds to minutes.

Wave-front sensing with time-of-flight phase diversity.

We present a new way to sense atmospheric wave-front phase distortion. Short collimated pulses of laser light at ~350nm are projected from a small auxilliary telescope. Rayleigh scattering from each pulse is recorded over a wide range of height through the main telescope aperture in a continuous sequence of fast video frames by a detector conjugate to mid-height.

Sensing wave-front amplitude and phase with phase diversity.

We show in benchtop experiments that wave-front phase estimation by phase diversity can be significantly improved by simultaneous amplitude estimation. Processing speed, which will be important for real-time wave-front control applications, can be enhanced by use of small-format detectors with pixels that do not fully sample the diffraction limit. Using an object-independent phase-diversity algorithm, we show that, for both pointlike and extended objects, the fidelity of the phase and amplitude estimates degrades gracefully, rather than catastrophically, as the sampling becomes coarser.

Blind deconvolution in optical diffusion tomography.

We use blind deconvolution methods in optical diffusion tomography to reconstruct images of objects imbedded in or located behind turbid media from continuous-wave measurements of the scattered light transmitted through the media. In particular, we use a blind deconvolution imaging algorithm to determine both a deblurred image of the object and the depth of the object inside the turbid medium. Preliminary results indicate that blind deconvolution produces better reconstructions than can be obtained using backpropagation techniques.