Audio Signal Analysis, Sound Design
The goal of Audio Signal Analysis, Sound Design is the integration of theoretical results of UNLocX into sound signal processing field. These results will yield a new point of view on localization for waveform systems in a signal/image processing perspective, together with explicit constructions of corresponding waveform systems, and efficient numerical algorithms and implementations. We will demonstrate the efficiency of these new methods for specific sound signal processing applications in the industry: acoustic signal analysis, sound design and real-time audio synthesis. This work package will lead to the proof of concept in the sound processing domain.
The aims are as following:
- Sound signal analysis: We shall develop methods for estimating physically meaningful information from signal representations optimally designed and implemented in UNLocX. The developments shall be done in the context of a limited number of specific industrial applications, such as car engine diagnosis from acoustic measurements, condition monitoring of rotating machines, tracking of partials/components/patterns, masking and sound signal coding and compression.
- Sound design: We aim at devising sound design systems that allow control over sound modification from a limited amount of parameters, signal modification (automatic separation of components, pitch shifting, time stretching, denoising, morphing, frequency masking effects...). These will exploit the optimal signal representations as well as the results obtained in audio signal analysis and the efficient algorithms developed in UNLocX.
- Real-time sound synthesis: We shall investigate the implementation of the above mentioned developments (parameter estimation, sound design) within real time systems. Such an evolution is indeed crucial in several industrial applications. The aim is to conform to real-time processing requirements that are used in the domain of realistic soundscape simulation.