We present a project concerning a method devoted to the learning of the phonetics of a given language. This system is written in hypertext and uses the function of a speech signal editor (Snorri). The learning concerns the pronunciation of vowels and of consonants in a reduced context, of syllables and words (study of coarticulatory phenomena), of prosody as well as an initiation to spontaneous speech.
Our method relies upon the comparison between the users pronunciation and the standard one (the "target"). To this purpose, the user has at his disposal: the play-back of his utterances and that of the target, their spectrographic representations with formant trackings, the projection of their formant values in the vocalic space, for consonants, the values of the main acoustic cues.
Snorri also displays the variation of the fundamental frequency (intonation) along the time. The energy curve and the durations of any part of the sentence is also available.
We will constitute a database containing coarticulatory and spontaneous speech phenomena. The user will be allowed to visualise and play back these phenomenons at will.
A book, written in hypertext, will explain to the user the main phonetic notions and the learning method.