Audio Signal Resampling by Weighted Interpolation for Low Delay

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current audio frequency signal resampling methods, such as polynomial interpolation, face limitations in reducing complexity and maintaining signal quality, especially for high-frequency undersampling, leading to significant algorithmic delays and artifacts like pre-echoes and temporal spreading.

Innovation Solution

A resampling method that calculates an interpolated sample value by taking a weighted average of possible interpolation values over multiple intervals, specifically using cubic interpolation across three intervals, reduces computational complexity while improving signal-to-noise ratio and maintaining signal quality.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If polynomial interpolation is used for resampling, then computational complexity is reduced and delay is minimized, but signal quality deteriorates due to aliasing artifacts and pre-echo effects

Engineering Contradiction:
Improvecomputational complexityVSAvoidsignal quality
Core Design Contradiction:
Device complexityVSReliability

Solution Approach 1:

The patent introduces an intermediary anti-aliasing filter between the polynomial interpolation stage and the output. This filter removes high-frequency aliasing components that would otherwise degrade signal quality, while the polynomial interpolation itself maintains low complexity and low delay. The intermediary filter selectively eliminates harmful frequency components without requiring a complete redesign of the interpolation algorithm.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The resampling process is segmented into distinct functional stages: first polynomial interpolation to generate intermediate samples with minimal delay, then anti-aliasing filtering to remove artifacts. This segmentation allows each stage to be optimized independently - the interpolation for speed and the filtering for quality - rather than requiring a single complex operation to handle both concerns.

Inventive Principle:
Principle #1Segmentation

2Reliability

If FIR filtering is used for resampling, then signal quality is improved with linear phase response, but algorithmic delay increases significantly

Engineering Contradiction:
Improvesignal qualityVSAvoidalgorithmic delay
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent extracts only the essential anti-aliasing function from a full FIR filtering operation. Instead of using a long impulse response FIR filter that would cause significant delay, the invention applies a simplified filtering operation that removes only the critical high-frequency aliasing components. This selective extraction of the anti-aliasing function maintains signal quality while minimizing time delay.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent applies partial filtering action - not the complete filtering that would be required for perfect signal reconstruction, but just enough filtering to remove the most harmful aliasing artifacts. This partial action is sufficient to maintain acceptable signal quality while avoiding the excessive delay that would result from complete FIR filtering.

Inventive Principle:
Principle #16Partial or excessive action

3Measurement precision

If high-order polynomial interpolation is used, then interpolation accuracy improves, but computational complexity increases

Engineering Contradiction:
Improveinterpolation accuracyVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent changes the parameter of polynomial order from high (which would provide better accuracy but increased complexity) to a moderate order (2 or 3). This parameter change is compensated by the subsequent anti-aliasing filter that removes the artifacts that would otherwise result from the lower-order interpolation. The parameter change reduces computational complexity while the filter maintains the effective accuracy.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3161659B1Resampling of an audio signal by interpolation for low-delay encoding/decoding
Publication Date: 2022.07.27 ORANGE SA
  • EP3161659B1 patent drawingFigure 1
  • EP3161659B1 patent drawingFigure 2
  • EP3161659B1 patent drawingFigure 3

AI summary

The invention relates to a method for resampling an audio-frequency signal in an audio-frequency signal encoding or decoding operation, the resampling being carried out by a method of interpolation of an order greater than one. The method is such that the interpolated samples are obtained (E702) by calculating a weighted average of possible interpolation values (E701) calculated over a plurality of intervals covering the time location of the sample to be interpolated. The invention also relates to a resampling device implementing the described method, and an encoder and decoder comprising at least one device.