Fast Adaptive Multichannel Equalization Identification and Control –(DANCE- FAME)
This subproject involves applications of Dynamic Acoustic Sensor Networks (DASN) for local control of noise, sound zones reproduction and generalized crosstalk cancellation on changing scenarios.
New sound field control strategies adapted to the environments where control or listening points may vary with time will be implemented and tested. Different control objectives will be considered at each listening or controlled area and they can dynamically change over time. It is intended that a different target sound field effect (cancellation, equalization, localized reproduction, etc.) could be selected by the user at each listening zone.
Classical techniques to characterize static indoor environments are not suitable enough due to the lack of knowledge about acoustic characteristics of the room, or because the environment and the position of microphones (listeners) and loudspeakers (sources) can be dynamic. Consequently, alternative approaches to spatial-temporal characterization of the room are necessary for indoor multi-zone and spatial audio applications in dynamic environments. Other aspects, such as information of the location of noise sources and obstacles and the coupling between the different loudspeakers and microphones of the Acoustic Sensor Network (ASN) need also to be addressed since they have great importance in the acoustic characterization of a room.
- Acoustic analysis of dynamic indoor environments. Classical techniques to characterize static indoor environments are not suitable enough due to the lack of knowledge about acoustic characteristics of the room, or because the environment and the position of microphones (listeners) and loudspeakers (sources) can be dynamic. Consequently, alternative approaches to spatial-temporal characterization of the room are necessary for indoor multi-zone and spatial audio applications in dynamic environments. Moreover, DANCE-FAME will search for algorithmic solutions that could eventually be used for network self-adapting of node placements, topology/clustering and/or node specific (or cluster specific) objectives.
-
Development of algorithms and strategies for active control and reproduction of sound DASNs. Distributed multichannel sound reproduction, and active control algorithms based on fast least squares and derived from the Kalman filter approach, will be developed and investigated under a common framework. They should exhibit fast adaptation and allow to be implemented over a distributed and heterogeneous network. Multiuser perceptual equalization methods to improve the listening experience in presence of undesired ambient noises will be also considered.
-
Implementation of applications for indoor dynamic real-life scenes. The final result will be demonstrated through the following testbeds:
-
-
A Personal audio system for indoor environments. This testbed will address applications of DASN for local control of noise, sound zones reproduction and generalized crosstalk cancellation on changing scenarios. The testbed will render personalized sound zones that, depending on their extension and the type of sound/noise involved, will be implemented by different algorithms. Examples are: watching TV and simultaneously listening to different languages in different positions, improving the listening experience in any room, tracking the listener over the home.
-
Massive multichannel noise reduction in open-plan offices. In this type of offices, workers usually get annoyed by the intelligible speech produced by other workers, which also means that speech privacy gets compromised. Other ambient noises can even contribute to the general discomfort. This testbed is intended to reduce this annoyance.
-
- González Salvador, Alberto
Head of the research group in Audio and Multimedia Digital Signal Processing (GTAC).
- de Diego Antón, Maria
Associate Professor
- Estreder Campos, Juan
PhD Student
- Ferrer Contreras , Miguel
Associate Professor
- Fuster Criado, Laura
Technical researcher
- García Mollá, Victor Manuel
Associate Professor
- Guerri Cebollada, Juan Carlos
Associate Professor
- Martinez Zaldivar, Francisco Jose
Associate Professor
- Moles Cases, Vicent
PhD. Stundent
- Piñero Sipán, Gema
Professor