Voice signal representations for reconstruction and modification

• Grant number: 07/08288-2
• Support Opportunities: Scholarships abroad - Research
• Start date: May 01, 2008
• End date: July 31, 2008
• Field of knowledge: Engineering - Electrical Engineering - Telecommunications
• Principal Investigator: Miguel Arjona Ramírez
• Grantee: Miguel Arjona Ramírez
• Host Investigator: Willem Bastiaan Kleijn
• Host Institution: Escola Politécnica (EP). Universidade de São Paulo (USP). São Paulo , SP, Brazil
• Institution abroad: KTH Royal Institute of Technology, Sweden

Abstract

The aim of this project is the enhancement of representations of the speech signal that decompose it into a linear predictive (LP) model and a residual signal in the form of a signal expansion on a warped time scale. The residual representation is continuously scalable in precision and forms an exact waveform description if all components are kept and an efficient parametric description if an optimal subset of the components is kept. Previously, the Signal Processing Laboratory (LPS) has proposed LP models that represent efficiently the spectrum of the slowly evolving component of the residual signal. New methods will be researched to make these models less sensitive to speech signal variations as well as new representations for the slowly evolving component of the residual signal will be explored taking comparatively models developed by KTH (Royal Institute of Technology) that are capable of perfect reconstruction (PR). (AU)