Complex Modulated Filterbank for Shorter Subband Audio Filters

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing filter banks in audio signal processing, particularly for HRTF applications, face challenges in achieving sufficient frequency resolution and efficient manipulation of time domain signals, with current methods often resulting in poor audio quality due to insufficient phase characteristics and computational inefficiencies.

Innovation Solution

A filter converter system that converts a given filter's impulse response into subband domain filters using a complex modulated filter bank, allowing for efficient filtering and manipulation of audio signals by downsampling and upsampling processes, reducing computational complexity and improving audio quality.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a complex-exponential modulated filter bank is used for spectral envelope adjustment, then frequency resolution is enhanced, but computational complexity increases

Engineering Contradiction:
Improvefrequency resolutionVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The audio signal is divided into multiple frequency subbands using a complex-exponential modulated filter bank. Each subband is processed independently with shorter filters, achieving high frequency resolution while reducing the computational burden on each individual filter operation.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transforms the filtering operation from the time domain to the frequency domain using a complex modulated filter bank. This dimensional transformation allows for more efficient processing by operating on frequency coefficients rather than time-domain samples, reducing overall computational complexity.

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

2Productivity

If simple gain modifications are applied in each subband, then processing efficiency is improved, but audio quality deteriorates due to insufficient phase characteristics

Engineering Contradiction:
Improveprocessing efficiencyVSAvoidaudio quality
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent applies different processing strategies to different parts of the frequency spectrum. In subbands where phase characteristics are critical for audio quality, full complex filtering is applied rather than simple gain modification. This local adaptation ensures high audio quality in critical regions while maintaining processing efficiency in less critical regions.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent modifies the filtering approach by changing from simple real-valued gain modifications to complex-valued filtering operations in specific subbands. This parameter change introduces proper phase characteristics where needed, improving audio quality while maintaining the efficiency benefits of subband processing.

Inventive Principle:
Principle #35Parameter changes

3Device complexity

If HRTF filtering is performed in the subband domain, then computational complexity is reduced, but filter conversion complexity increases

Engineering Contradiction:
Improvecomputational complexityVSAvoidfilter conversion complexity
Core Design Contradiction:
Device complexityVSEase of manufacture

Solution Approach 1:

The patent pre-computes and stores the filter conversion matrices that map from time-domain HRTF filters to subband-domain filters. This preliminary action allows the complex conversion operation to be performed once offline, rather than in real-time during audio processing, reducing the computational burden during actual operation.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent creates a copied representation of the HRTF filters in the subband domain through a systematic transformation process. Instead of repeatedly performing complex convolutions in the time domain, the filters are copied and transformed into the subband domain once, enabling efficient repeated use during audio rendering.

Inventive Principle:
Principle #26Copying

Data Source

PatentEP4178110B1Efficient filtering with a complex modulated filterbank
Publication Date: 2024.04.24 DOLBY INTERNATIONAL AB
  • EP4178110B1 patent drawingFigure 1a~1c
  • EP4178110B1 patent drawingFigure 1d~1e
  • EP4178110B1 patent drawingFigure 2~4

AI summary

A filter converter (104) for converting an impulse response signal indicative of an amplitude/frequency filter characteristic in a time domain into filter definition signal, the converter comprising a complex modulated filter bank (301) for filtering the impulse response signal to obtain a plurality of complex valued subband signals forming the filter definition signal, wherein each complex valued subband signal corresponds to an impulse response for a subband filter (102). At least one complex subband signal comprises at least two different non-vanishing values and each complex valued subband signal is shorter than the impulse response signal. The impulse response signal is a time domain FIR (Finite Impulse Response) filter. The complex modulated filter bank (301) is adapted to output L complex valued subband signals and the complex modulated filter bank (301) is adapted for providing the L complex valued subband signals downsampled by a factor L.