COAST: Context-aware pervasive speech recognition system

Context-aware applications adapt their behavior to the user current situation. This paper presents a new architecture named COAST (Context-aware Speech to text translator). Reducing user interaction and selecting the best classifier based on contexts are the primary objectives in COAST and user's privacy rules can be applied too. The contexts are categorized in two sets: system-contexts and classification contexts. The system contexts adapt systems behaviors. The Classification contexts guide COAST to select current classifiers and modify some of them. COAST can work without server to enable autonomic behavior. Clients can connect to peers to achieve more advantages such as: fault-tolerance feature, with severs connection, achieving more contexts from the other clients' resources.