As the authors of Distant Speech Recognition note, automatic speech recognition is the key enabling technology that will permit natural interaction between humans and intelligent machines. Core speech recognition technology has developed over the past decade in domains such as office dictation and interactive voice response systems to the point that it is now commonplace for customers to encounter automated speech-based intelligent agents that handle at least the initial part of a user query for airline flight information, technical support, ticketing services, etc. While these limited-domain applications have been reasonably successful in reducing the costs associated with handling telephone inquiries, their fragility with respect to acoustical variability is illustrated by the difficulties that are experienced when users interact with the systems using speakerphone input. As time goes by, we will come to expect the range of natural human-machine dialog to grow to include seamless and productive interactions in contexts such as humanoid robotic butlers in our living rooms, information kiosks in large and reverberant public spaces, as well as intelligent agents in automobiles while traveling at highway speeds in the presence of multiple sources of noise. Nevertheless, this vision cannot be fulfilled until we are able to overcome the shortcomings of present speech recognition technology that are observed when speech is recorded at a distance from the speaker.