Audio Encoding Noise Reduction via MDCT Block Switching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current audio encoding methods, such as MPEG AAC, face inefficiencies in noise reduction due to high calculation costs and discontinuous noise reduction across varying frame sizes, leading to suboptimal noise reduction performance.
Innovation Solution
The proposed method employs modified discrete cosine transformation (MDCT) to convert audio signals into long or short blocks, applying non-linear multi-band spectral subtraction and spectral reduction based on signal-to-noise ratio (SNR) and perceptual sound quality curves, with block switching to adapt to variable frame lengths, reducing noise while maintaining performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If non-linear multi-band spectral subtraction is applied to long blocks and spectral reduction to short blocks, then noise reduction performance is improved, but calculation complexity increases
Solution Approach 1:
The audio signal is divided into long blocks (1024 points) and short blocks (128 points) based on signal characteristics. Different noise reduction algorithms are applied to each block type: non-linear multi-band spectral subtraction for long blocks and spectral reduction for short blocks. This segmentation allows optimization of calculation complexity for each block type while maintaining overall noise reduction performance.
Solution Approach 2:
Different noise reduction techniques are applied locally to different block types based on their specific characteristics. Long blocks receive non-linear multi-band spectral subtraction which is more computationally intensive but provides better noise reduction for stationary signals. Short blocks receive simpler spectral reduction suitable for transient signals, reducing unnecessary calculation complexity while maintaining effectiveness.
2Adaptability or versatility
If block switching is implemented to handle variable frame lengths, then adaptability to different audio signals is improved, but device complexity increases
Solution Approach 1:
The noise reduction apparatus dynamically switches between long block and short block processing modes based on the characteristics of the input audio signal. This dynamic adaptation allows the system to optimize performance for different signal types (stationary vs. transient) while managing computational resources efficiently. The block switching mechanism enables the system to adapt to variable frame lengths without requiring completely separate processing paths.
3Reliability
If spectral subtraction with perceptual sound quality curves is performed, then noise reduction effectiveness is improved, but calculation time increases
Solution Approach 1:
Perceptual sound quality curves and subtraction coefficients are pre-calculated and stored based on different signal-to-noise ratio conditions. During real-time noise reduction, the system retrieves these pre-computed values and applies them to the audio signal, significantly reducing calculation time while maintaining noise reduction effectiveness. This preliminary preparation of reference data allows the system to achieve accurate noise reduction without performing complex calculations during signal processing.
Data Source
AI summary
A method and apparatus for audio signal encoding for noise reduction are provided. The method includes: receiving an audio signal and performing modified discrete cosine transformation (MDCT) on the audio signal to convert the audio signal into a long block or a short block; reducing noise included in the audio signal in accordance with the long block or the short block; and performing advanced audio coding (AAC) on the long block or the short block in which noise is reduced.