A method prepares voice guidance instructions for an individual, formulated in the most natural fashion possible. The method includes:—determining, by a computer, a path to be followed, —acquiring, by an image acquisition unit, at least one image of the environment of the individual, —processing the image in order to detect at least one object therein and in order to characterise the object, and —preparing a voice guidance instruction supplying the user with a piece of information for carrying out a manoeuvre in order to follow the path. The method also includes determining a level of complexity of the manoeuvre, and, in the preparation step, the voice guidance instruction is formulated using an indication deduced from the characterisation of the object only if the level of complexity of the manoeuvre is greater than a first threshold.