Extracting and Re-rendering Structured Auditory Scenes from Field Recordings

Emmanuel Gallo (1, 2) , Nicolas Tsingos (1)

1 REVES - Rendering and virtual environments with sound
2 CSTB - Centre Scientifique et Technique du Bâtiment

We present an approach to automatically extract and re-render a structured auditory scene from field recordings obtained with a small set of microphones, freely positioned in the environment. From the recordings and the calibrated position of the microphones, the 3D location of various auditory events can be estimated together with their corresponding content. This structured description is reproduction-setup independent. We propose solutions to classify foreground, well-localized sounds and more diffuse background ambiance and adapt our rendering strategy accordingly. Warping the original recordings during playback allows for simulating smooth changes in the listening point or position of sources. Comparisons to reference binaural and B-format recordings show that our approach achieves good spatial rendering while remaining independent of the reproduction setup and offering extended authoring capabilities.

Mots clés: Intelligent Audio Environments

Domaines: Synthèse d'image et réalité virtuelle [cs.GR]

Emmanuel Gallo, Nicolas Tsingos. Extracting and Re-rendering Structured Auditory Scenes from Field Recordings. 30th International Conference: Intelligent Audio Environments (AES 2007), Mar 2007, Saariselkä, Finland. pp.11. ⟨inria-00606798⟩