The following report describes the research that I have performed or directed over the last 20 years covering the period from my PhD thesis (A. Röbel 1993) until today. After a short introduction discussing the evolution and invariants of the research orientation in chapter 1, the early research activities, before joining IRCAM, are presented in a very short summary in chapter 2. The main part of the report will then describe my research on spectral domain analysis, modelling, and transformation of sound that has been conducted at IRCAM during the last 13 years. First, the specificities of the research environment at IRCAM and the general objectives of my research activities will be briefly discussed in chapter 3. Signal analysis and signal transformation using the short time Fourier transform (STFT) and extensions of the phase vocoder algorithm are covered in chapter 4. Ad- vanced methods for STFT based signal processing that are currently developed and not yet used in practical applications will be presented in chapter 5. The three methods covered include: signal adaptive resolution, texture transformation and source separation. My research on sinusoidal modelling taking into account non-stationary sinusoidal components is presented in chapter 6. Research results related to source-filter models, notably spectral envelope estimation and instrument timbre modelling, are described in chapter 7. The research related to the problem of fundamental frequency estimation for sounds containing single or multiple quasi harmonic sources are described in chapter 8. The following chapter 9 discusses my research on speech signal processing, covering the three research problems: glottal pulse parameter estimation, voice conversion, and singing synthesis. Activities related to the development of signal processing libraries are described in chapter 10. In the final chapter of this document (chapter 11) I will discuss the four major directions that I expect to be important for my research during the next five years.