We have lately developed a non-invasive multi-sensor acquisition set – the hyper-helmet – for rare songs recording in an intangible cultural heritage safeguarding perspective. In this presentation, we take advantage of this articulatory sensing system to study and test a new nasality index. The helmet’s acoustic microphone and nasal piezoelectric accelerometer are used to calculate an oral/nasal rms ratio. An ElectroGlottoGraph instrument is the mean to estimate the voicing selector parameter. In addition, a non-intrusive tongue imaging sensor (an ultrasonic probe) and a lips movement camera are backups for articulatory and nasality qualitative interpretation. A software has been developed for synchronous acquisition of all sensors and it is been used to record an English corpus interpreted by a native English-speaking Canadian mid-age man. Multiple tests have been held to verify numerous nasality theories. Some results are shown in this presentation.
Prochains événements
Voir la liste d'événementsSoutenance d'Andrés Lara
Du signal au geste, du geste au signal : classification automatique et modélisation bayésienne des rhotiques en français et en anglais américain


