Archontis Politis, Sakari Tervo, and Ville PulkkiCOMPASS: Coding and Multidirectional Parameterization of Ambisonic Sound ScenesCompanion page for a paper in the 2018 IEEE International Conference on Acoustics, Speech and Signal Processing, Calgary, Alberta, Canada, April 15–20, 2018AbstractCurrent methods for immersive playback of spatial sound content aim at flexibility in terms of encoding and decoding, abstracting the two from the recording or playback setup. Ambisonics constitute such a method, that is however signal-independent, and at low spatial resolutions fails to provide appropriate spatialization cues to the listener, with potential severe colouration effects and localization ambiguity. We present a new signal-dependent method for parametric analysis and synthesis of ambisonic sound scenes that takes advantage of the flexibility of Ambisonics as a spatial audio format, while improving reproduction. The proposed approach considers a more general acoustic model than previous proposals, with multiple source signals and a non isotropic ambient component. According to a listening test using headphones, the method is perceived closer to binaural reference sound scenes than ambisonic playback. PaperThe paper has been submitted and is under review - if accepted it will be available in the IEEE publications website.Listening test sound scenesThe following samples correspond to the five sound scenes used in the MUSHRA listening test. The reference test signals correspond to direct binaural rendering using the same non-individualized set of HRTFs for all subjects. The same sound scenes were also encoded to first- and third-order Ambisonics, to be processed by the proposed method. The cases to be tested were first- and third-order ambisonic rendering for headphones, and through the proposed COMPASS method, all using the same set of HRTFs as the reference. Dry scenes correspond to anechoic conditions with multiple sources, while in the reverberant ones reverberation was added using the image-source method, with frequency-dependent reverberation times and air absorption modeling. Three tests are conducted: a) the overall test, in which the samples are compared without modifications, b) the spatial test, in which the root mean square spectrum of the binaural signals of the test cases is matched to the one of the reference case, so that timbral differences are minimized and spatial differences and processing artifacts become dominant, and c) the timbral test in which the reference case is replicated and its RMS spectrum is matched to each of the test cases, so that the generated signals exemplify only timbral differences between the rendering methods. Overall TestBand Dry
Band Rev
Speakers Dry
Soundscape Rev
Orchestra Dry
Spatial TestBand Dry
Band Rev
Speakers Dry
Soundscape Rev
Orchestra Dry
Timbral TestBand Dry
Band Rev
Speakers Dry
Soundscape Rev
Orchestra Dry
| ||