A method and system for dynamically controlling a voice-processing mechanism in a voice command platform. The platform receives a specification during a voice command session with a user and responsively sets a mode of operation of the voice-processing mechanism. The platform can receive the specification from various sources, such as a user (e.g., as a voice command), a stored user profile record, a voice command application and/or another stored table or data source. Example modes of operation include (i) use of a designated text-to-speech engine from among multiple text-to-speech engines, (ii) use of a designated voice prompt store from among multiple voice prompt stores, (iii) muting of a speech recognition engine, and (iv) selection of a secondary phoneme dictionary for use in recognizing phonemes in incoming speech signals.