Correcting incomprehensible synthetic speech

A method and system for speech synthesis. In a text-to-speech system comprising a text input is received and the text entry is processed using a processor of the system to synthetic speech is determined by that it is unintelligible. Text input is reprocessed to subsequent synthetic speech and output via a loudspeaker to a user to correct the unintelligible synthesized speech. In one embodiment, it can be stated that the synthetic speech is not understandable by the intelligibility of synthetic speech is predicted and it is determined that the predicted intelligibility is lower than a minimum threshold value. In another embodiment, it can be stated that the synthetic speech is not understandable by the synthetic voice is output from the speaker to the user and is received from the user an indication that the synthetic speech is not understandable.