1 / 22

Study of Neural Correlates of Mandarin Tonal Production with Neural Network Model

Study of Neural Correlates of Mandarin Tonal Production with Neural Network Model . Chao-Min Wu* Tao-Wei Wang . Department of Electrical Engineering, National Central University, Jhongli 320, Taiwan, ROC. Received 6 May 2011; Accepted 15 July 2011; doi : 10.5405/jmbe.934 .

erwin
Download Presentation

Study of Neural Correlates of Mandarin Tonal Production with Neural Network Model

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Study of Neural Correlates of Mandarin Tonal Production with Neural Network Model Chao-Min Wu* Tao-Wei Wang Department of Electrical Engineering, National Central University, Jhongli 320, Taiwan, ROC Received 6 May 2011; Accepted 15 July 2011; doi: 10.5405/jmbe.934 Chairman:Hung-Chi YangPresenter: Yue-Fong Guo Advisor: Dr. Yeou-Jiunn ChenDate: 2012.12.26

  2. Outline • Introduction • Overview of the DIVA model • Method • Results • Discussion • Conclusion

  3. Introduction • Population ages • Communicative disorders • Improve the diagnosis and treatment of speech problems • Analyzing Mandarin speech production • Neural correlates of Mandarin tonal production

  4. Introduction • Physiological model • To Examine speech articulation • A neural-network-based direction into DIVA models (Directions Into Velocities Articulator ) • Simulate neural correlates of speech production • To determine the neural correlates of Mandarin tonal production

  5. Introduction • The DIVA model • Originally design • Fixed preset pitch • Tonal production was not considered • The modified DIVA model • Produce four types of Mandarin tones • Simulate brain activity regions

  6. Overview of the DIVA model • Adaptive neural network model • The steps • A word or a syllable input • Generate articulatory movement commands • Shape of the vocal tract required to produce • Speech synthesizer produce speech sound • Produces a sequence of number • Represent the brain activity regions and levels

  7. Overview of the DIVA model

  8. Method • DIVA model • Mandarin tonal production is described and incorporated • Simulating brain activity are presented

  9. Method • Mandarin tonal production • Chao proposed a five-point-scale

  10. Method • Pitch scaling function • Approximated with the fourth-order polynomials • Multiplied by the pitch periods of the first tone

  11. Method • To modify the motor commands • Corresponding articulator to generate • The tonal speech • The corresponding brain activity regions

  12. Method • Simulation of brain activity • Baseline condition • Speaking condition • Corresponding brain areas

  13. Method • In the first simulation • Given vowel with different tones(/a/, /á/, /ã/, and /à/) • Were analyzed • Verify whether capable of tone production • In the second simulation • The brain activity regions of two different vowels with a given tone(/a/-/u/) • The difference of the corresponding vowel brain activity regions • Determine whether the original function maintained

  14. Method • In the third simulation • Give vowel with different tones • A comparison among brain activity regions

  15. Results • In the first experiment • According to the first two formant frequencies • The produced f0 contours • Capable of Mandarin tonal production

  16. Results • In the second experiment • motor cortex, pre-motor cortex, auditory cortex, SMA, and cerebellum • (a) the vowel /a/ and (b) the vowel /u/

  17. Results • Difference between the production of different vowels • Difference in the activation in the lip and laryngeal areas and the SSM area

  18. Results • In the third experiment • Difference in the larynx area and the somatosensory cortex

  19. Discussion • Previous study find the tone-relate activity regions • Prefrontal cortex • Pitch judgment tasks • Left middle temporal • Word-level comprehension • Right inferior frontal gyri • Precentralgyri

  20. Discussion • The DIVA model simulating brain activity • Motor cortex • Auditory • Somatosensory • Cerebellum • The differences between simulation and study

  21. Conclusion • Because these regions are not included in the DIVA model. • Future study • Focus on the tone-related brain region • Needed to include these region in the DIVA model • To investigate the role • The frontal and temporal lobes

  22. The end

More Related