A speech coding method using an anthropomorphic and acoustic approach

Usually, speech synthesis technologies use signal processing methods such as unit concatenation techniques. However, speech synthesis can also be carried on by an anthropomorphic approach which consists in modeling the whole speech production system starting from the description of the geometry of the vocal tract (area function). This human - like approach can be considered as an interesting way to code speech signal to reduce transmission rates for telecommunications. We will present a new anthropomorphic model consisted by an analog mechanic - acoustic model of the vocal cords and the vocal tract, coupled with the Distinctive Region Model (DRM) to build a complete synthesizer. We will show first results on Vietnamese speech language and discuss the usability of this model for speech coding.

@InternationalConference{NC07,
  author       = {NGUYEN, V. and CASTELLI, E.},
  title        = {A speech coding method using an anthropomorphic and acoustic approach},
  booktitle    = {The proceeding of the 10th conference of oriental chapter of the international coordinating committee on speech database and speech I/O system assessment (O-COCOSDA 2007)},
  pages        = {155--159},
  month        = {December},
  year         = {2007},
  url          = {/2007/NC07},
}