Descripción del proyecto
INTELLIGENT SPATIAL AUDIO: ANALYSIS AND MONITORING (ISLA-ALAMO) IS THE COORDINATING SUBPROJECT OF THE JOINT ACTION PROPOSAL ISLA (INTELLIGENT SPATIAL AUDIO: MACHINE-LEARNING ASSISTED SYNTHESIS AND MONITORING),TRADITIONAL AUDIO SIGNAL PROCESSING AND ACOUSTIC-BASED APPLICATIONS HAVE BEEN DEEPLY IMPACTED BY THE RAPID GROWTH IN THE LAST FEW YEARS OF MACHINE LEARNING, HIGH-PERFORMANCE SINGLE-BOARD COMPUTERS AND HIGH-BANDWIDTH WIRELESS COMMUNICATION NETWORKS, THESE TECHNOLOGIES HAVE POSED IMPORTANT CHANGES IN THE WAY THAT EMERGING ACOUSTIC SIGNAL PROCESSING APPLICATIONS ARE BEING CONCEIVED, WHILE CLASSIC ACOUSTIC PROCESSING SYSTEMS HAVE BEEN DEVELOPED TAKING INTO ACCOUNT HIGH-THROUGHPUT COMPUTING PLATFORMS EQUIPPED WITH EXPENSIVE MULTICHANNEL AUDIO INTERFACES, THE INTERNET-OF-THINGS PARADIGM IS DEMANDING THE USE OF MORE FLEXIBLE, ENERGY-EFFICIENT AND SMART SYSTEMS, ISLA-ALAMO INTENDS TO BRING ALL THESE TECHNOLOGIES TOGETHER WITH THE AIM OF PUSHING AUDIO SIGNAL PROCESSING FORWARD TOWARDS A NEW ERA WHERE A HIGH-LEVEL UNDERSTANDING OF THE ACOUSTIC ENVIRONMENT IS EXPECTED TO RESULT IN INNOVATIVE CONTEXT-AWARE SYSTEMS, THE RESEARCH CONDUCTED IS THEREFORE AIMED AT TURNING SOUND INTO A SENSORING MODALITY WITH HIGH EXPLOITATION POTENTIAL, AMONG THE CHALLENGES TO BE CONSIDERED ARE THE LOCATION OF MULTIPLE SOURCES WITH DISTRIBUTED MICROPHONE NETWORKS, SYNCHRONIZATION AND SELF-CALIBRATION OF FIXED AND MOBILE NODES WITH WIRELESS CONNECTIVITY, MULTIMODAL TECHNIQUES FOR THE DESIGN OF SUPERVISED LEARNING ALGORITHMS, THE IDENTIFICATION AND TRACKING OF SOUND OBJECTS OR THE CREATION OF PSYCHOACOUSTIC MODELS FOR THE ESTIMATION OF DESCRIPTIVE PARAMETERS BEYOND THE SOUND PRESSURE LEVEL, THE ABOVE CONTEXT FINDS A PARTICULAR AND APPEALING APPLICATION FRAMEWORK IN SPATIAL AUDIO, I,E, THE MANIPULATION OF AUDIO SIGNALS WITH THE AIM OF CAPTURING AND/OR RECREATING THE SPATIAL IMPRESSION OF SOUND, DISTRIBUTED MICROPHONE NETWORKS COMBINED WITH MACHINE LEARNING ARE EXPECTED TO BE KEY ELEMENTS FOR EXTRACTING HIGH-LEVEL DESCRIPTIONS OF THE SOUNDSCAPE: IDENTIFYING, LOCALIZING AND TRACKING THE RELEVANT SOUND OBJECTS MAKING UP A SOUND SCENE, THE INFORMATION GATHERED BY SUCH AN ANALYSIS AND MONITORING SYSTEM WILL IN TURN MAKE POSSIBLE THE RECREATION OF SOUNDSCAPES WITH CONTEXT-AWARE INTERACTIVE AND INTELLIGENT FEATURES, IT IS THEREFORE EXPECTED THAT THE NOVELTY OF THIS APPROACH, TOGETHER WITH THE COORDINATED WORK OF EXPERTS WITH A LONG TRAJECTORY IN SPATIAL AUDIO, WILL LEAD TO AN AMBITIOUS PROPOSAL WITH A HIGH EXPLOITATION POTENTIAL, SONIDO\APRENDIZAJE MÁQUINA\PROCESADO DE AUDIO\REDES DE SENSORES\ACÚSTICA