Réverbération hybride et réponses impulsionnelles spatialisées

Edit

Content:
<img src="/media/uploads/recherche/projets/reverberationhybride.png" width="448" height="350" style="float: left; margin-right: 10px;" /> Most artificial reverberators use either feedback delay networks (FDN) or convolution reverb. These approaches differ in their computation cost, the “authenticity” of the results, the flexibility of use and adaptability vis-.-vis the system used for audio rendering. For several years, the Acoustic and Cognitive Spaces team has explored a “hybrid” approach. Using measured room impulse responses (RIR), the early reflections are reproduced by convolution while the later part of the reverberation is synthesized by a FDN that approximates the energy decay relief (EDR) of the RIR. The transition from one mode to the other occurs at the so-called “mixing time” and the FDN is calibrated to guarantee the continuity of the EDR, assuring audio rendering that is perceptively indistinguishable from the original IRR. This hybrid technique has been extended to spatial room impulse responses (SRIR), measured with spherical microphone arrays (see photo) in order to recreate the spatial distribution of the room reflections. For several years, the Acoustic and Cognitive Spaces team and its collaborators (artists in residence, sound engineers, etc.) have been collecting directional impulse responses (DRIR) in a variety of spaces (open air, concert halls, churches, archaeological sites, etc.). Initiated as part of the artistic research residency of Pedro Garcia-Velazquez and Augustin Muller, the library includes, to date, some 40 acoustic spaces. In order to make it available to the production department and, ultimately, to the general public, the various measurement campaigns have been annotated (positions of sources and receivers, architectural typology, etc.) and standardized (storage format, coordinate system, file naming, etc.). A demonstrator has been developed to 'navigate' in this catalog, to represent schematically the sound scene and to display the different associated annotation fields. IRCAM's Team : <a href="https://www.ircam.fr/recherche/equipes-recherche/eac">Acoustic and Cognitive Spaces</a> Image : Assortment of remarkable sites used for 3D acoustic fingerprints © P. Garcia-Velasquez, A. Muller

Most artificial reverberators use either feedback delay networks (FDN) or convolution reverb. These approaches differ in their computation cost, the “authenticity” of the results, the flexibility of use and adaptability vis-.-vis the system used for audio rendering. For several years, the Acoustic and Cognitive Spaces team has explored a “hybrid” approach. Using measured room impulse responses (RIR), the early reflections are reproduced by convolution while the later part of the reverberation is synthesized by a FDN that approximates the energy decay relief (EDR) of the RIR. The transition from one mode to the other occurs at the so-called “mixing time” and the FDN is calibrated to guarantee the continuity of the EDR, assuring audio rendering that is perceptively indistinguishable from the original IRR.

This hybrid technique has been extended to spatial room impulse responses (SRIR), measured with spherical microphone arrays (see photo) in order to recreate the spatial distribution of the room reflections.

For several years, the Acoustic and Cognitive Spaces team and its collaborators (artists in residence, sound engineers, etc.) have been collecting directional impulse responses (DRIR) in a variety of spaces (open air, concert halls, churches, archaeological sites, etc.). Initiated as part of the artistic research residency of Pedro Garcia-Velazquez and Augustin Muller, the library includes, to date, some 40 acoustic spaces. In order to make it available to the production department and, ultimately, to the general public, the various measurement campaigns have been annotated (positions of sources and receivers, architectural typology, etc.) and standardized (storage format, coordinate system, file naming, etc.). A demonstrator has been developed to 'navigate' in this catalog, to represent schematically the sound scene and to display the different associated annotation fields.

IRCAM's Team : Acoustic and Cognitive Spaces

Edit