Adaptive Sub-Band Audio Filtering for FIR Resolution Limits

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing audio signal processing systems using FIR filters require long filter lengths to achieve sufficient frequency resolution, leading to high computational effort and memory requirements due to the characteristics of the human auditory system and typical listening rooms, which is inefficient.

Innovation Solution

Implementing an adaptive sub-band filtering system that divides the audio signal into multiple sub-bands, using a combination of analysis and synthesis filter banks with sub-band FIR filters and adaptive tap and delay assignment algorithms to optimize filter coefficients and delays, allowing for efficient implementation with reduced computational and memory demands.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If long FIR filter lengths are used to achieve sufficient frequency resolution, then audio quality is improved, but computational effort and memory requirements increase significantly

Engineering Contradiction:
Improvefrequency resolutionVSAvoidcomputational efficiency
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The audio signal is divided into multiple frequency sub-bands using analysis filter banks. Each sub-band is processed independently with shorter FIR filters, and the results are recombined using synthesis filter banks. This segmentation allows achieving sufficient frequency resolution in each sub-band with computationally efficient short filters, avoiding the need for a single long filter that would process the entire frequency spectrum.

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If long FIR filter lengths are used to achieve sufficient frequency resolution, then audio quality is improved, but memory requirements increase significantly

Engineering Contradiction:
Improvefrequency resolutionVSAvoidmemory requirements
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The filter bank structure divides the computational task across multiple short filters operating on different frequency sub-bands. Each sub-band filter requires minimal memory for its coefficients, whereas a single long filter covering the entire spectrum would require substantial memory. The segmented approach thus dramatically reduces total memory requirements while maintaining frequency resolution through the combined effect of multiple specialized filters.

Inventive Principle:
Principle #1Segmentation

3Productivity

If sub-band filtering with short filters is used, then computational efficiency is improved, but frequency resolution may be insufficient

Engineering Contradiction:
Improvecomputational efficiencyVSAvoidfrequency resolution
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The solution transitions from a single-dimensional approach (one long filter processing the entire spectrum) to a multi-dimensional approach where multiple short filters process different frequency dimensions simultaneously. The analysis filter banks separate the spectrum into distinct sub-bands, each processed by optimized short filters, and the synthesis filter banks recombine them. This dimensional transformation allows short filters to achieve the effective frequency resolution of a long filter by operating in the frequency domain rather than time domain.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

4Adaptability or versatility

If adaptive tap and delay assignment algorithms are implemented, then filter optimization is improved, but device complexity increases

Engineering Contradiction:
Improvefilter optimizationVSAvoidsystem complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system employs adaptive algorithms that dynamically adjust filter coefficients and delay parameters based on the characteristics of the audio signal and desired frequency response. Rather than using fixed, manually tuned parameters, the adaptive tap and delay assignment algorithms automatically optimize filter performance in real-time, allowing the system to adapt to different audio content and achieve optimal frequency resolution and computational efficiency for each specific application scenario.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentUS8879747B2Adaptive filtering system
Publication Date: 2014.11.04 HARMAN BECKER AUTOMOTIVE SYST GMBH
  • US8879747B2 patent drawing
  • US8879747B2 patent drawing
  • US8879747B2 patent drawing

AI summary

An audio system with at least one audio channel may include a digital audio processor in which at least one digital filter is implemented for each channel. The digital filter of each channel may include: an analysis filter bank configured to receive a broad-band input audio signal and divide the input audio signal into a plurality of sub-bands, a sub-band filter for each sub-band, and a synthesis filter bank configured to receive the filtered sub-band signals and combine them for providing a broad-band output audio signal. A delay is associated with each sub-band signal, the delay of one of the sub-band signals being applied to the broad-band input audio signal upstream of the analysis filter bank and the residual delays being applied to the remaining sub-band signals downstream of the analysis filter bank.