Downloads

This section of Nonlinear Statistical Modeling of Speech includes software that has been developed at CAVs. We have two main distributions that we work with. Our Java Applet is one of these distributions. We have updated the applet with our "Linear Prediction" method and our "Particle Filtering" method. Be encouraged to revisit our site in order to keep up with our progress.

Our other distribution is a research environment that includes a generalized hierarchical Viterbi search-based decoder and a network trainer. We use this environment to construct utilities, involving the use of support vector machines (SVMs) and relevance vector machines (RVMs). This environment should only be used by serious speech and signal processing researchers and can be found at the link below.

Available Downloads

• Java Applet: We have extended our popular pattern recognition applet to include our first implementation of emerging techniques in nonlinear time series modeling known as particle filtering. Revisit our site for updates to this applet.



• Rapid Prototyping : We have our first implementation of emerging techniques in nonlinear time series modeling known as Kalman filtering, Lyapunov exponents, and particle filtering. Click on the links above to download the matlab code.



• The Production System: A research environment that includes a generalized hierarchical Viterbi search-based decoder and a network trainer. Recommended for serious speech and signal processing researchers. There is also a downloadable version of the software documentation at the above link [tar.gz].



• SVM/RVM Training: A tutorial that provides an overview of the control flow for our hybrid system approach to implementing risk minimization-based approaches such as SVMs and RVMs.