Filtering directive invoking vocal utterances

A method comprising:

• receiving, from a user, entered text that specifies a candidate resource name for a first resource of a computer system;
• responsive to determining that a speech rendering of the entered text sounds confusingly similar to a speech rendering of an existing resource name communicating, to the user, information indicating that the speech rendering of the entered text sounds confusingly similar to a speech rendering of an existing resource name;
• and
• receiving second entered text from the user.

A method comprising:

• receiving a candidate audio data set with the candidate audio data set including:
• (i) a candidate text proposed for association with a candidate text based command, and (ii) audio data corresponding to a candidate speech rendering of the candidate text;
• determining that speech recognition software is likely to misidentify utterances of the candidate text as corresponding to a text based command other than the candidate text based command;
• and
• performing an action in dependence on the determining.