1. Björkner, Eva Researcher, Doctoral Student Address Helsinki University of Technology Laboratory of Acoustics and Audio Signal Processing P.O. Box 3000 FIN-02015 HUT HOARSE sept ’03 Swedish Singer Teacher in singing and music

2. MR and Acoustic Study of Throaty Voice Quality †*Eva Björkner, † Johan Sundberg † Department of Speech Music Hearing, KTH, Stockholm, Sweden * Laboratory of Acoustics and Audio Signal Processing, Helsinki University of Technology, Finland HOARSE sept ’03

3. What is throaty voice quality? How and why does it sound as it does? The purpose of the present investigation was to further analyse acoustic and articulatory characteristics of throaty voice quality. The investigation focuses on the two lowest formant frequencies and area functions based on MR data from a male speaker contrasting throaty and normal/habitual voice quality in four vowels HOARSE sept ’03 AUDIO data Standard text read by 1 male speaker. a. with habitual voice b. with throaty voice Listening test =>vowels [a, ae, i, u] most throaty

4. F1 & F2 for Habitual and Throaty version HOARSE sept ’03 Formant frequencies after inverse filtering pressure signal

5. Habitual versus Throaty Quality HOARSE sept ’03 F1 & F2 F3 & F4

6. imaging MR imaging HOARSE sept ’03

7. The same 4 vowels were pronounced for about 15 seconds in both qualities, while MR-images were shot of 14 sections from the larynx to the nose tip Teeth don´t show in MR images MR analysis of male subject HOARSE sept ’03 Vocal tract contour approximated by polygon Determination of area, center of gravity, length co-ordinate Center of gravity Area = 869 mm 2

8. The estimation of vocal tract length was not trivial. Section one, closest to the glottis, was located at an unknown distance above the glottis level. When constructing the area functions this distance was provisionally assumed to be 1 cm in all vowels. As the location of the various sections were basically fixed relative to the body, the first section would have varied relative to the glottis if the larynx position was changed.  A rise of the larynx would shorten the vocal tract length, eventhough this was not evident from the MR images.  Also lip protrution and spreading caused similar problems as no information on larynx height and lip conditions was available. Tricky estimations

9. Normal Throaty : closer to glottis => smaller piriform sinuses Piriform sinuses Section 1: fixed Differences in larynx height between the versions

10. i a ae u Area functions Narrowing Expansion

11. Conclusions Formants: Higher F1, lower F2 in front vowels, lower F4 Vocal tract: - narrow lower pharynx - expansion near the middle of the vocal tract - higher larynx position The pharyngeal narrowing produces an increase of F1 in the vowels /a, ae, i, u/ while the effects on F2 vary. HOARSE sept ’03

12. - professor Didier Demolin, Laboratoire de Phonologie Université Libre de Bruxelles, and his associates for making the MR imaging possible - professor Anne-Maria Laukkanen, Tampere University for running the listening test during her visit to the department of Speech Music Hearing, KTH. This investigation was produced within the HOARSE Network project and was partly carried out at the Laboratory of Acoustics and Audio Signal Processing, Helsinki University of Technology. Thanks to: HOARSE sept ’03Eva Björkner

13. This work as been presented at: VOQUAL’03: VOICE QUALITY: FUNCTIONS, ANALYSIS AND SYNTHESIS GENEVA - AUGUST 27-29, 2003 ISCA Tutorial and Research Workshop International Speech Communication Association The Voice Foundations 32 nd The Voice Foundations 32 nd Annual meeting in Philadelphia, USA, 4 - 8 june 2003