Generalized Envelope Matching for Time-Scale Modification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing time-domain time scale modification methods, such as synchronous overlap-and-add, require high computational resources due to cross-correlation and normalization processes, limiting their efficiency in resource-limited applications.
Innovation Solution
A generalized envelope-matching time scale modification method that uses the n most significant bits for higher accuracy and employs a fixed-size cross-correlation buffer to eliminate normalization within the search loop, leveraging fast/parallel shift and multiply-and-accumulate instructions for improved precision and reduced computational cost.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If synchronous overlap-and-add method with cross-correlation and normalization processes is used, then time scale modification quality is improved, but computational cost increases
Solution Approach 1:
The patent extracts only the most significant bits (MSBs) of the signal values for cross-correlation computation, discarding less significant bits. This extraction approach maintains the essential envelope information needed for accurate overlap point detection while significantly reducing the computational burden of processing full-precision signal values throughout the search loop.
Solution Approach 2:
The patent changes the parameter representation from full-precision signal values to truncated values using only the n most significant bits. This parameter transformation reduces the computational complexity of cross-correlation and normalization operations while preserving the critical envelope matching information required for high-quality time scale modification.
2Power
If envelope-matching time scale modification using only the sign of samples is used, then computational cost is reduced, but accuracy decreases
Solution Approach 1:
The patent generalizes the envelope-matching approach by changing from using only the sign bit (1 bit) to using the n most significant bits (where n>1). This parameter change increases the resolution of envelope detection and improves accuracy while still maintaining computational efficiency by processing only the most significant portions of the signal values.
Solution Approach 2:
The patent applies partial action by processing only the most significant bits of the signal rather than complete precision values. This partial processing approach provides sufficient accuracy for envelope matching while avoiding the excessive computational cost of processing full-precision data throughout the entire search loop.
3Measurement precision
If variable-size cross-correlation buffer with normalization is used, then search accuracy is improved, but device complexity increases
Solution Approach 1:
The patent performs preliminary normalization by using a fixed-size cross-correlation buffer, which eliminates the need for dynamic normalization calculations within the search loop. This preliminary structuring of the buffer simplifies the computational requirements and reduces device complexity while maintaining search accuracy through the fixed buffer dimensions.
Solution Approach 2:
Instead of using a variable-size buffer that requires dynamic normalization, the patent inverts the approach by using a fixed-size buffer that inherently simplifies the normalization process. This inversion from variable to fixed sizing reduces the computational complexity and device requirements while preserving the essential search functionality.
Data Source
AI summary
A time-domain time-scale modification method based on the synchronous overlap-and-add method consists of a generalization of the envelope-matching time-scale modification method. The cross-correlation function employs n most significant bits rather than merely the sign bit of the prior envelope matching method. This provides higher accuracy for n>1. A fixed-size cross-correlation buffer is employed to eliminate the need for normalization inside the search loop. This invention makes full use of fast/parallel shift and multiply-and-accumulate (MAC) instructions of current digital signal processors to become at the same time faster and more precise than envelope-matching time-domain time-scale modification.