Quaternion Domain Quantization for Ambisonic Audio Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing ambisonic encoding methods, such as multi-mono and quantization of rotation matrices using Euler angles, result in spatial distortions and artifacts at low bitrates, particularly in immersive audio applications.
Innovation Solution
Encoding and decoding ambisonic signals using quaternion domain quantization, where one quaternion component is forced to be positive, simplifying conversions and reducing complexity by using spherical coordinates and optimized quantization methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If multi-mono encoding is used for ambisonic signals, then implementation is simple, but spatial distortions and artifacts occur at low bitrates
Solution Approach 1:
The patent transforms the encoding approach by changing the parameter representation from independent mono channels to a unified ambisonic format with correlation preservation. This involves using spherical harmonics expansion and optimized quantization of rotation matrices, which maintains spatial relationships between channels while reducing bitrate requirements.
Solution Approach 2:
The patent replaces the simple parallel mono encoding mechanism with a more sophisticated ambisonic encoding system that uses mathematical transformations (spherical harmonics, rotation matrices). This substitution enables better spatial accuracy by modeling the acoustic field relationships rather than treating channels independently.
2Ease of manufacture
If rotation matrices are quantized using Euler angles, then conversion is straightforward, but spatial precision deteriorates at low bitrates
Solution Approach 1:
The patent changes the quantization parameters from Euler angles to a optimized representation based on spherical harmonics coefficients and rotation matrix elements. This parameter transformation maintains the mathematical properties of rotation matrices while enabling more efficient and precise quantization at lower bitrates, preserving spatial accuracy.
Solution Approach 2:
The patent moves the quantization approach from traditional Euler angle parameterization to a higher-dimensional representation using spherical harmonics coefficients. This dimensional change allows for better preservation of spatial relationships by capturing the full rotational information in a form that is more amenable to optimized quantization.
3Measurement precision
If high bitrate is used for ambisonic encoding, then spatial quality is maintained, but processing power and memory requirements increase
Solution Approach 1:
The patent transforms the data representation to use spherical harmonics coefficients and optimized rotation matrix quantization, which reduces the number of parameters that need to be transmitted and processed. This parameter optimization maintains spatial quality by preserving the essential acoustic field information while reducing bitrate and associated processing requirements.
Solution Approach 2:
The patent extracts and transmits only the essential parameters needed for spatial reconstruction (spherical harmonics coefficients and quantized rotation matrices) rather than transmitting all raw channel data. This extraction approach maintains spatial quality by preserving the critical spatial information while reducing the overall data volume and processing burden.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
The invention relates to a method for encoding a multichannel sound signal, comprising forming a transformation matrix in the form of a rotation matrix to be applied to the input signals, quantifying the rotation matrix, encoding the transformed signals after application of the rotation matrix and in which the quantifying of the rotation matrix comprises the following operations: - converting (E300) the rotation matrix in the quaternion domain with at least one first quaternion; - forcing (E320) said first quaternion to have a positive component; - converting the at least one first quaternion into spherical coordinates, one of the spherical coordinates being associated with the forced positive component of the first quaternion; - quantifying (E330) the resulting rotating spherical coordinates, the spherical coordinate associated with the forced positive component of the first quaternion being quantified in a half-length interval. The invention also relates to a corresponding decoding method, an encoding device and a decoding device.