When comparing phrases recognized by the speech recognition with the string I typed in, the result was always false. The reason is the following: If a speech phrase is recognized it is stored in the variable with lower case letters even if the phrase was written in the wait for speech command with a mix of lower case and capital letters.