Accounting for Microprosody in Modeling Intonation

Peter Birkholz; Xinyu Zhang

ICASSP 2020accepted0 citations

Accounting for Microprosody in Modeling Intonation

Abstract

Intonation models are often used for the generation of fundamental frequency (f 0 ) contours in speech synthesis. Current intonation models only represent the intentional f 0 components that are related to the phonological structure of the utterance. However, natural speech also contains non-intentional microvariations of f 0 , which are usually not accounted for. Here, we derived models for two forms of microvariations: the drop in f 0 during voiced obstruents, and the increased f 0 at the onset of vowels following voiceless obstruents. These models were applied to remove the microvariations of f 0 in a database of natural speech before the f 0 contours were reproduced with the Target Approximation Model. The previously removed microvariations were then superimposed on the modeled f 0 contours. The resulting model f 0 contours were significantly more similar to the original (natural) f 0 contours than model contours that did not account for the mi-crovariations. This approach might improve f 0 modeling in future parametric speech synthesizers.

BibTeX

@inproceedings{icassp2020_accountingformic,
  title = {Accounting for Microprosody in Modeling Intonation},
  author = {Peter Birkholz and Xinyu Zhang},
  booktitle = {ICASSP 2020},
  year = {2020}
}