Say It To See It: A Speech-Based Immersive Model Retrieval System

Abstract:Traditionally, 3D models used within a virtual reality application must reside locally on the system in order to be properly loaded and displayed. These models are often manually created using traditional 3D design applications such as Autodesk products or Trimble SketchUp. Within the past five to 10 years, web-based repositories of 3D models, such as the Trimble 3D Warehouse, have become available for sharing 3D models with other users around the world. PC speech recognition capabilities have also become much more commonplace and accessible to developers, particularly with the advent of the Microsoft Kinect. We’ve developed a method to perform a search of an online model repository, the Trimble 3D Warehouse, from within our fully immersive CAVE by using speech. Search results are displayed within the CAVE, and the user can select a model he or she wishes to download and see drop into the CAVE. The newly acquired model can then be instantly manipulated and placed within the user’s current scene at a desired location.