Stereo Audio Encoding with Shared Excitation and Spatial Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current stereo encoding techniques generate excessive coding information due to separate adaptive and fixed codebooks for each channel, leading to increased bit rates and potential deterioration of speech quality in decoded signals.
Innovation Solution
A stereo encoding apparatus that corrects similarity between channel signals using spatial information analysis and encoding, allowing for shared excitation and reduced coding information by encoding spatial information differences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If separate adaptive and fixed codebooks are used for each channel in stereo encoding, then speech quality of each channel is maintained, but the amount of coding information increases
Solution Approach 1:
The patent merges the adaptive codebook and fixed codebook into a unified codebook structure that serves both channels. Instead of maintaining separate codebooks for left and right channels, a single shared codebook is used to generate excitation signals for both channels, thereby reducing the amount of coding information while preserving speech quality through shared spectral characteristics.
Solution Approach 2:
The unified codebook serves multiple functions: it generates excitation signals for both left and right channels simultaneously, and provides a common basis for cross-channel prediction. This multi-functional approach eliminates the need for separate codebooks while maintaining the ability to preserve speech quality across both channels.
2Reliability
If separate codebooks are used for each channel, then channel-specific speech characteristics are preserved, but encoding bit rate increases
Solution Approach 1:
The patent combines separate channel-specific codebooks into a single unified codebook that serves both channels. This merging reduces the encoding bit rate by eliminating redundant codebook data while preserving speech quality through the application of cross-channel prediction techniques that exploit the correlation between left and right channel signals.
Solution Approach 2:
The patent introduces cross-channel prediction as an intermediary mechanism that bridges the gap between using a unified codebook and preserving channel-specific characteristics. By predicting one channel from the other and applying the prediction residual to the unified codebook excitation, the system maintains channel-specific speech qualities while benefiting from the reduced bit rate of shared coding.
3Productivity
If cross-channel prediction is applied with separate codebooks, then coding efficiency is improved, but the complexity of the encoding system increases
Solution Approach 1:
The patent merges the codebook structures to reduce system complexity. By using a single unified codebook instead of separate codebooks for each channel, the patent simplifies the encoding system architecture while maintaining coding efficiency through cross-channel prediction that operates on the shared codebook excitation signals.
Data Source
AI summary
A stereo audio encoding apparatus capable of preventing degradation of the sound quality of a decoded signal, while reducing the encoding bit rate. A spatial information analyzer analyzes spatial information for each of a L channel signal and an R channel signal. A similarity raiser corrects, based on an analysis result of the spatial information analyzer, a difference between the spatial information of the L channel signal and that of the R channel signal, to raise a similarity between the L and R channel signals. A channel signal encoder uses a sound source common to the two channels to encode the L and R channel signals as raised in similarity and output the resultant single encoded information. A spatial information encoder encodes the analysis result of the spatial information analyzer to output the resultant encoded information.