|
The acoustics of soprano unintelligibility| old_uid | 12128 |
|---|
| title | The acoustics of soprano unintelligibility |
|---|
| start_date | 2016/09/30 |
|---|
| schedule | 09h30-10h30 |
|---|
| online | no |
|---|
| details | Ce séminaire (en libre accès) aura lieu dans le cadre de l'édition 2016 du colloque PAC, intitulée "Les mélodies de l'anglais", qui se tiendra au Laboratoire Parole et Langage à Aix-en-Provence, du 29 septembre au 1er octobre 2016. |
|---|
| summary | Soprano singing is notoriously difficult to comprehend. The reasons for that are only partially understood, and in fact there has also been a recent demonstration that the loss of intelligibility may have been overstated. In this talk I will report an experiment in which CV syllables sung by a soprano are heard by listeners with some phonetic training. The singer recorded syllables on an arpeggio, with the stimuli used in the experiment ranging from ‘middle’ A (A4, ≈ 440 Hz) to A5 (≈ 880 Hz). The vowels were a subset – [i, ?, a, ?, ?, u] – of the phonetic ‘cardinal vowels’, plus [?], and listeners had to identify the vowel in a sung syllable, and also to say whether or not there was a consonant preceding the vowel in each syllable. In brief, there was a strong tendency for vowels to converge perceptually on an open [a]-like vowel as pitch increased, and for the initial consonant percept to be lost. However, some vowel discrimination was preserved above a pitch where a loss of vowel quality would be predicted by a formant-based model of vowel perception – that is, one which relies on inference of the resonances of the vocal tract from the available harmonics. An explanation for this supererogatory distinctiveness will be put forward. The implications of the findings will be discussed, and our results compared with those recent results showing unexpected survival of phonetic quality at extreme high pitch. |
|---|
| responsibles | Contact |
|---|
| |
|