In:
Journal of the Acoustical Society of America, Acoustical Society of America (ASA), Vol. 140, No. 4_Supplement ( 2016-10-01), p. 3219-3219
Abstract:
While real-time magnetic resonance imaging (rt-MRI) provides high spatiotemporal resolution for speech research, the associated audio is noisy, presenting a challenge for research on the relationship between articulation and the acoustic signal and solving the articulatory inversion problem. Using state-of-the-art denoising methods, the current study denoised rt-MRI audio associated with nasal and oral French vowels produced by one speaker, and extracted F1-3 from the midpoints of each vowel for /a, o, e/ and their nasal counterparts. Oblique images were taken of the velopharyngeal port at 25 frames/second, and average pixel intensity (API) in the velopharyngeal region was taken from images corresponding to the vowel midpoint. General additive models showed a significant relationship between API and F1 for oral and nasal vowels. (Lower API indicates a wider velopharyngeal opening.) F1 of /ã/ was lower than /a/, while F1 was higher in the nasal /õ/ and /ẽ/ than their oral counterparts, all of which are expected effects of naso-pharyngeal coupling. These results show that the acoustic recordings produced during rt-MR imaging can be used to explore the relationship between articulation and acoustics, and can give insight into the articulatory effects of nasality, something difficult to understand through acoustics alone.
Type of Medium:
Online Resource
ISSN:
0001-4966
,
1520-8524
Language:
English
Publisher:
Acoustical Society of America (ASA)
Publication Date:
2016
detail.hit.zdb_id:
1461063-2