If the user transposes and then maps freqDegreeToKey with round set to 'nearest', the the user gets the wrong answer. Unclear if the other roundings are effected. Fix for the nearest is to sort the normalised array of semitones.