Segmentation of speech on phonetic elements for systems of speech information protection