IMPROVED SPEECH TO TEXT METHOD AND SYSTEM

The invention concerns a method for converting speech to text for a speech communication system (SCS), comprising the steps of: capturing a speech; converting the speech to a text by a speech-to-text (STT) engine, and making the text available, wherein the STT engine is configured to perform the steps of: a) converting the speech to a set of text fragments using a generic speech recognition algorithm; b) for each text fragment: checking if one or more words of the text fragment comprise a meaning corresponding to one of a predefined set of one or more semantic options; c) if one or more words of the text fragment corresponds to a semantic option: comparing said one or more words with a set of entries in a database corresponding to said semantic option and computing a trust level score for each of said entries, thereby obtaining an optimal entry having the highest-trust level score, d) correcting each text fragment by replacing the one or more words which comprise a meaning corresponding to one of a predefined set of one or more semantic options with said optimal entry, thereby obtaining a set of corrected text fragments and e) converting the set of corrected text fragments to a presentable text, wherein the database corresponding to the semantic option is adaptable, and the database corresponding to the semantic option is configured to be updated on the basis of a current state of the SCS.