Recording speech movements using magnetometry: one laboratory’s experience.
| Number | 894 |
|---|---|
| Year | 1993 |
| Drawer | 16 |
| Entry Date | 07/15/1998 |
| Authors | Löfqvist, Anders, Gracco, Vincent L., and Nye, Patrick W. |
| Contact | LOFQUIST@YALEHASK (Bitnet) -or- LOFQUIST%YALEHASK@VENUS.YCC.YALE.EDU |
| Publication | Proceedings of the ACCOR Workshop on Electromagnetic Articulography in Phonetic Research, 91-104. München, Germany. |
| url | http://www.haskins.yale.edu/Reprints/HL0894.pdf |
| Abstract | For studies of speech production, safe and reliable methods for transducing articulatory movements are necessary. Over the years, several different methods have become available and been applied to speech articulation, such as x-ray, x-ray microbeam, ultrasound, stain gages, optoelectronic techniques, and Magentic Resonance imaging. While x-ray provides an image of the whole vocal tract, its inherent safety problems (arising from subject exposure to ionizing radiation) impose limitations on the amount of data that can be recorded. The x-ray microbeam reduces the radiation exposure, but at present no system is available for general use. Strain gages and optoelectronic procedures offer excellent resolution and can provide three-dimensional records but only on lip and jaw movements. Ultrasound is limited in the number of articulators that can be tracked simultaneously. Magnetic Resonance Imaging provides excellent images of the whole vocal tract but, at present, image acquisition times are too long to permit movement tracking; this may well change in the near future, however. The development of electromagnetic transduction as an alternate measurement technique (Schönle, 1988 Perkell, Cohen, Svirsky, Matthies, Garabieta, and Jackson, 1992) offers a number of advantages compared to other methods, provided that proper care is taken during data collection - many receivers can be tracked simultaneously, the spatial and temporal resolution is good, the absolute positions of receivers can be measured, and all the necessary signal processing can be automated and simplified. The present paper presents a number of our experiences at Haskins Laboratories in using an electromagnetic articulometer. Some of our current thinking about, and approaches, to processing magnetometer signals will be illustrated with data on articulatory movements in VCV sequences, with particular emphasis on tongue movements. Since several of our approaches are still exploratory, they are likely to change in the future. Other practical and theoretical problems that we have encountered as discuss by Gracco and Nye (1993). |
| Notes |
Search Publications