An algorithm for speech parameter generation from continuous mixture HMMs with dynamic features | Synapse