QMF Filter Compression to Reduce HRTF Processing Complexity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing filter conversion techniques face substantial computational complexity, particularly with filters having long impulse responses in the time-domain, such as HRTF filters, which are used for multi-channel audio processing, leading to inefficiencies in computational resources.
Innovation Solution
A filter system comprising a filter converter and a filter compressor that converts time-domain filters into a complex QMF representation, selectively setting lower-valued filter impulse responses to zero to reduce computational complexity while maintaining audio quality, using a mask generator to identify and preserve higher-valued coefficients.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If HRTF filters with long impulse responses are used for multi-channel audio processing, then audio quality is improved, but computational complexity increases substantially
Solution Approach 1:
The patent extracts only the essential information from the long HRTF filter impulse responses by converting them to QMF domain coefficients and applying a mask that retains only the most significant coefficients. This extraction process removes redundant data while preserving the critical audio quality characteristics, thereby reducing computational complexity without sacrificing audio fidelity.
Solution Approach 2:
The patent transforms the filter representation from the time domain to the QMF (Quadrature Mirror Filter) domain, changing the parameter space in which the filters are expressed. This parameter transformation allows for more efficient representation and processing, reducing computational complexity while maintaining the ability to reproduce the original audio quality when needed.
2Productivity
If QMF domain representation is used for filtering, then filtering efficiency is improved, but computational complexity remains substantial for long impulse response filters
Solution Approach 1:
The patent extracts only the essential information from the long HRTF filter impulse responses by converting them to QMF domain coefficients and applying a mask that retains only the most significant coefficients. This extraction process removes redundant data while preserving the critical audio quality characteristics, thereby reducing computational complexity without sacrificing audio fidelity.
Solution Approach 2:
The patent applies partial action by using a mask to retain only a subset of the QMF coefficients rather than processing all coefficients. By keeping only the most significant coefficients (those above a certain threshold or within a specific range), the system achieves sufficient filtering efficiency with reduced computational effort, avoiding the need to process every coefficient fully.
3Measurement precision
If all filter impulse response values are processed, then filtering accuracy is maintained, but computational resources are wasted on low-valued coefficients
Solution Approach 1:
The patent applies local quality by differentiating between significant and insignificant filter coefficients. Instead of treating all coefficients uniformly, the system applies a mask that identifies and retains only the locally significant coefficients (those with higher values) while discarding or simplifying the processing of low-valued coefficients. This localized approach maintains filtering accuracy where it matters most while conserving computational resources.
Solution Approach 2:
The patent discards low-valued filter coefficients that contribute minimally to the overall filtering accuracy, and recovers or reconstructs the full filter response only when necessary for maintaining precision. The masking mechanism allows the system to discard insignificant data during normal operation, recovering full accuracy only when the masked coefficients would be needed for high-fidelity processing.
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
A filter system comprises a filter converter (101) and a filter compressor (102) for generating compressed subband filter impulse responses from input subband filter impulse responses corresponding to subbands, which comprise filter impulse response values at filter taps. The filter compressor comprises a processor (820) for examining the filter impulse response values from at least two input subband filter input responses to find filter impulse response values having higher values and at least one filter impulse response value having a value being lower than the higher values, and a filter impulse response constructor (305) for constructing the compressed subband filter impulse responses using the filter impulse response values having the higher values, wherein the compressed subband filter impulse responses do not include filter impulse response values corresponding to filter taps of the at least one filter impulse response value having the lower value or comprise zero-valued values corresponding to filter taps of the at least one filter impulse response value having the lower value.