Audio Signal Processing with Dynamic Linear Predictive Coefficient Order
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audio signal processing technologies face challenges in applying a consistent sampling rate across different bandwidths, leading to inefficient bit allocation and reduced coding efficiency due to fixed orders of linear-predictive coefficients.
Innovation Solution
An apparatus and method that dynamically adjust the order of linear-predictive coefficients based on the bandwidth and coding mode of the audio signal, allowing for the same sampling rate to be applied across varying bandwidths, and utilizing recurring properties to optimize bit allocation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If different sampling rates are applied for narrow band and wide band audio signals, then encoding accuracy is improved, but device complexity increases due to requiring different cores
Solution Approach 1:
The patent applies a single sampling rate universally for both narrow band and wide band audio signals, allowing one core to handle multiple bandwidth scenarios. This eliminates the need for separate cores for different sampling rates, thereby reducing device complexity while maintaining encoding accuracy through adaptive order adjustment of linear-predictive coefficients
Solution Approach 2:
Instead of changing the sampling rate parameter, the patent changes the order parameter of linear-predictive coefficients to adapt to different bandwidths. This parameter substitution allows the system to maintain a fixed sampling rate while still achieving accurate encoding for various bandwidth conditions
2Device complexity
If a single sampling rate is applied for all bandwidths, then device complexity is reduced, but bit efficiency deteriorates due to fixed order of linear-predictive coefficients
Solution Approach 1:
The patent makes the order of linear-predictive coefficients dynamic by adapting it according to the bandwidth of the audio signal. For narrow band signals, a lower order is used, while for wide band signals, a higher order is used. This dynamic adjustment optimizes bit efficiency while maintaining a single sampling rate
Solution Approach 2:
The patent changes the order parameter of linear-predictive coefficients based on bandwidth conditions to optimize bit allocation. This parameter adaptation ensures that bits are allocated efficiently according to the actual signal characteristics, preventing waste while maintaining coding accuracy
3Ease of manufacture
If fixed order of linear-predictive coefficients is used, then implementation simplicity is maintained, but bit allocation efficiency deteriorates for narrow band signals
Solution Approach 1:
The patent introduces dynamic adjustment of the order parameter based on bandwidth detection. The system monitors the bandwidth of the audio signal and adjusts the order of linear-predictive coefficients accordingly. This dynamic approach optimizes bit allocation efficiency while maintaining relatively simple implementation through standard signal processing techniques
Data Source
AI summary
The present invention relates to a method for processing an audio signal, comprising: determining bandwidth information indicating to which of a plurality of bands the current frame corresponds; determining information on the order corresponding to the present frame on the basis of the bandwidth information; performing a linear predictive analysis of the present frame to generate a first set linear predictive transform coefficient of a first order; performing a vector quantization on the first set linear predictive coefficient to generate a first index; performing a linear predictive analysis of the current frame to generate a second set linear predictive transform coefficient of a second order in accordance with the information on the order; and performing a vector quantization on a second set difference by using the first set index and the second set linear predictive transform coefficient, when the second set linear predictive coefficient is generated.