Implementation of cross-domain, context-sensitive speech analysis