Inter-channel prediction reference signal selection for stereo audio
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing inter-channel prediction (ICP) methods for stereo speech coding face challenges in achieving satisfactory performance, especially when channel correlation is low, leading to high prediction errors in the low-frequency band, and introduce algorithmic delay due to the use of transform coders for wideband audio.
Innovation Solution
A coding apparatus and decoding apparatus that utilize a monaural signal generation section, side residual signal acquisition, monaural residual signal acquisition, spectrum division, selection of optimal reference signals, and inter-channel prediction analysis to improve ICP performance by selecting the best reference signal among candidates for prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If transform coders are used for wideband audio coding, then audio quality is improved, but algorithmic delay is introduced
Solution Approach 1:
The audio signal is divided into different frequency bands (low band and high band) which are processed separately. This segmentation allows the system to use different coding approaches for different bands, reducing overall algorithmic delay while maintaining audio quality.
Solution Approach 2:
Different coding methods are applied to different frequency bands based on their specific characteristics. The low band uses one coding approach while the high band uses another, optimizing both quality and delay performance for each local frequency region.
2Device complexity
If conventional ICP methods are used with low channel correlation, then coding complexity is reduced, but prediction error increases in low-frequency band
Solution Approach 1:
The patent transitions from time-domain processing to frequency-domain processing by applying MDCT transformation. This dimensional change allows the system to exploit spectral properties and inter-channel correlations in the frequency domain, improving prediction accuracy for low-frequency components while maintaining manageable coding complexity.
Solution Approach 2:
The system changes the representation parameters from time-domain samples to frequency-domain MDCT coefficients. This parameter transformation enables better exploitation of spectral characteristics and inter-channel relationships, particularly improving low-frequency prediction accuracy.
Data Source
AI summary
An encoder improves inter-channel prediction (ICP) performance in scalable stereo sound encoding using an ICP. In the encoder, ICP analysis units use, as reference signal candidates, a frequency coefficient in the low-band portion of a side residual signal, a frequency coefficient in each sub-band portion of a monaural residual signal, and a frequency coefficient in the low-band portion of the monaural residual signal, respectively, and perform an ICP analysis between the these respective candidates and a frequency coefficient in each sub-band portion of the side residual signal to generate first, second, and third ICP coefficients. A selection unit selects an optimum reference signal from among the reference signal candidates by checking the relationship between the respective reference signal candidates and the frequency coefficient in each sub-band portion of the side residual signal and outputs, to an ICP parameter quantization unit, a reference signal ID indicating the selected reference signal and an ICP coefficient corresponding to the reference signal.