An engineering blog for engineers working in Audiovisuals, Telecommunications, Electronics, ICT, Computer Engineering, Multimedia and Telematics.

11 June 2010 | Posted by Redacción Ingeniería

New DEA presented: Expressive synthesis based on HNM speech signal parametrisation

A new DEA (Advanced Studies Diploma) was successfully presented last June 8th 2010 within the research group on Media Technologies, in the area of multimodal processing, by the PhD candidate Angel Calzada, here there is a summary of his research: This work is about expressive synthesis based on HNM (Harmonic plus Noise Model) speech signal parametrisation. The advantages of this model rely in the flexibility for modifying the speech signals. A review of the state of art in expressive speech synthesis has been carried out, pointing out the advantages and drawbacks of the most used expressive synthesis techniques. Also some voice conversion methods adapted to expressive style conversion have been reviewed. Results and conclusions obtained from multiple studies carried out along this last year have been exposed and commented. Basically, the exposed work is focused on the application of the HNM model for voice manipulation and expressive synthesis with Voice Quality (VoQ). Finally the work concludes pointing out some research lines to be explored with the actual implementation of the HNM system.

Share