Cascade of asymmetric resonators with fast-acting compression

This past spring I took a trip to the US to visit several research groups and attend some conferences. During the trip, I visited the Machine Perception group at Google Research and got involved in an open source project implementing a C++ version of an auditory image model. The model was designed by Richard Lyon, a researcher and engineer who has spent many years investigating human auditory perception and developing mathematical models of the specific processes and transformations which occur as sound is converted by the ear from air vibrations into neurally encoded information. Such models can be used in the context of machine perception, a branch of artificial intelligence that aims to give rich perceptual abilities to computing applications, including the ability to understand speech but also to ‘make sense’ the rich sound environments which humans effortlessly navigate during everyday life. You can read more about this type of approaches in this article by Richard Lyon. The current version of the model can be found on Github.

Leave a Reply

You must be logged in to post a comment.