Abstract

ECMA-407, the first 3D audio standard worldwide, introduces a new concept of static models to lower bitrate coding, which may be equally applied with channels, channels and objects and Higher Order Ambisonics (HOA). Static models may either operate in time domain or in frequency domain and allow transporting highly complex 3D audio content up to NHK 22.2 with no or very little side information, as would be necessary with dynamic models, generally referred to as parametric coding. Static models in time domain, when calibrated with statistical means, would require extensive computational complexity and large amount of data. A new approach, based on David Hilbert’s extensive studies on invariant theory, contrarily introduces a new “non-random” concept with Gaussian processes. Apart from an already published solution to the problem, an alternative solution is given here for the first time, entirely based on Hilbert’s ingenious construction of „kanonische Nullformen”. ECMA-407 is compliant with waveform preserving base audio codecs and non-waveform preserving audio codecs such as USAC and HE-AAC v2. Dataflow-programming by means of RVC-CAL will furthermore establish a programming environment, which is particularly apt for ECMA-407. The development, spatial tuning and testing environment for ECMA-407 at McGill University, conformant to ITU-R Recommendation BS.1116-1, is described together with a short description of subjective performance with PCM and with non-tuned and tuned non-waveform preserving based audio codecs.

Details

Actions