John Bridle

TL;DR: In this paper, a speech trigger threshold can be dynamically adjusted in response to perceived events (e.g., events indicating a user may be more or less likely to initiate speech interactions, events indicating that a trigger may be difficult to detect, events indicated a trigger was missed, etc.), thereby minimizing both missed triggers and false positive triggering events.

TL;DR: In this paper, an energy change of each frame making up signals including speech and non-speech signals is detected and a speech segment corresponding to frames that include only speech signals from among the frames based on the detected energy change.

TL;DR: A novel approach to the problems of topic and speaker identification that makes use of large-vocabulary continuous speech recognition and some empirical results on topic and Speaker identification that have been obtained on the extensive Switchboard corpus of telephone conversations are presented.

TL;DR: A straightforward primary detector is described and variations that result in very useful reductions in computation (or increased accuracy for the same computation) are explored that can be reduced by a factor of six while maintaining the same accuracy.

TL;DR: A probabilistic neural network model of acoustic trajectories, trajectory RNADE, is introduced, able to capture the dependencies between acoustic features conditioned on the phonetic labels in order to obtain high quality synthetic speech.