Fixed-Length Spectral Encoding via Log Envelope Smoothing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing signal processing techniques for time series signals, such as audio signals, face issues with distortion and compatibility during compression encoding due to errors in spectral envelope codes, leading to incorrect decoding and output signals.
Innovation Solution
The method involves obtaining a log spectral envelope sequence and using it to smooth and quantize frequency spectral values, allowing for fixed-length code assignment and reversible operations to maintain signal integrity and compression efficiency even with errors in spectral envelope codes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If variable-length coding is used based on spectral envelope to achieve efficient compression, then compression efficiency is improved, but decoding reliability deteriorates when errors occur in spectral envelope codes
Solution Approach 1:
The patent changes the parameter of code length from variable to fixed by introducing smoothing processing. The spectral envelope is used to smooth the frequency spectrum before quantization, which allows the code length to be determined in advance as a fixed value, thereby improving decoding reliability while maintaining compression efficiency through the spectral envelope information
Solution Approach 2:
The patent applies preliminary smoothing action to the frequency spectrum using the spectral envelope before quantization and coding. This preliminary processing modifies the spectrum characteristics in advance, enabling fixed-length coding and ensuring that even if errors occur in spectral envelope codes, the decoding can proceed with the correct number of samples
2Reliability
If fixed-length coding is used to ensure decoding reliability, then decoding reliability is improved, but compression efficiency deteriorates
Solution Approach 1:
The patent changes the parameter of code length from variable to fixed by introducing smoothing processing. The spectral envelope is used to smooth the frequency spectrum before quantization, which allows the code length to be determined in advance as a fixed value, thereby improving decoding reliability while maintaining compression efficiency through the spectral envelope information
Solution Approach 2:
The patent introduces spectral envelope as an intermediary element that mediates between the frequency spectrum and the quantized spectrum. This intermediary provides information about the spectral characteristics, enabling both efficient compression (by exploiting spectral redundancy) and reliable decoding (by enabling fixed-length coding through smoothing)
3Quantity of substance
If spectral envelope coding is used to reduce code amount, then code amount is reduced, but signal fidelity deteriorates when errors occur in spectral envelope codes
Solution Approach 1:
The patent applies preliminary smoothing action to the frequency spectrum using the spectral envelope before quantization and coding. This preliminary processing modifies the spectrum characteristics in advance, enabling fixed-length coding and ensuring that even if errors occur in spectral envelope codes, the decoding can proceed with the correct number of samples
Solution Approach 2:
The patent provides beforehand cushioning by using the smoothed spectrum as a reference that is less sensitive to errors in spectral envelope codes. The smoothing processing creates a more robust representation that cushions against the effects of potential errors, maintaining signal fidelity even when code amount is reduced through efficient spectral envelope coding
Data Source
Figure 1A~1B
Figure 2A~2B
Figure 3A~3C
AI summary
A log spectral envelope sequence L0, L1, ..., LN-1 and an envelope code for the log spectral envelope sequence L0, L1, ..., LN-1 are obtained. The log spectral envelope sequence L0, L1, ..., LN-1 is an integer value sequence corresponding to binary logarithms of respective sample values of a spectral envelope sequence and is an integer value sequence whose total sum is 0. For a quantized spectral sequence ^X0, ^X1, ..., ^XN-1, a smoothed spectral sequence ∼X0, ∼X1, ..., ∼XN-1 is obtained by: for ^Xk with Lk being a positive value, adopting ^Xk with Lk digits from its least significant digit removed as ∼Xk; for ^Xk with Lk being a negative value, adopting ^Xk with -Lk digits added to its least significant digit in accordance with a predefined rule as ∼Xk; and when Lk is 0, adopting ^Xk as ∼Xk. The respective samples of the smoothed spectral sequence ∼X0, ∼X1, ..., ∼XN-1 are then encoded with a fixed code length to obtain a signal code.