Sub-Band Resonance Attenuation for Low-Latency Music Processing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Live sound reinforcement and broadcasting face challenges in quickly and efficiently removing unwanted resonances and tonal imbalances due to high computational demands from low-frequency processing, especially with high sampling rates and large FFT sizes, which are resource-intensive and costly.

Innovation Solution

A method that subsamples frequency sub-bands of a sound signal to reduce computational resources, involving filtering, resonance analysis, convolution, upsampling, and combining sub-band blocks to form a result signal, allowing for efficient attenuation of resonances with reduced processing power requirements.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If high sampling rates (44.1 kHz to 192 kHz) are used for digital sound processing, then sound quality and processing flexibility are improved, but computational cost and processing power requirements increase significantly

Engineering Contradiction:
Improvesound qualityVSAvoidprocessing power
Core Design Contradiction:
Measurement precisionVSPower

Solution Approach 1:

The patent divides the frequency spectrum into multiple sub-bands using filter banks, processing each sub-band separately at reduced sampling rates. This segmentation allows high-quality processing where needed while reducing overall computational load by applying lower sampling rates to sub-bands that don't require full-resolution processing.

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If large FFT sizes are used for low-frequency processing (80 Hz to 260 Hz), then processing accuracy is improved, but computational resources and processing time increase

Engineering Contradiction:
Improveprocessing accuracyVSAvoidprocessing speed
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent segments low-frequency signals into specific sub-bands and processes them separately. By isolating low-frequency content into dedicated sub-bands, the system can apply appropriate FFT sizes tailored to each frequency range, avoiding the need for uniformly large FFT sizes across the entire spectrum and thus reducing overall computational burden.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent dynamically adjusts FFT size and sampling rate parameters based on the specific frequency band being processed. For low-frequency sub-bands, appropriate FFT sizes are selected to achieve necessary accuracy, while for higher frequencies, smaller FFT sizes suffice, optimizing the balance between accuracy and processing speed across different frequency ranges.

Inventive Principle:
Principle #35Parameter changes

3Measurement precision

If full-resolution processing is applied to all frequency bands, then sound quality is maintained, but computational cost and latency increase

Engineering Contradiction:
Improvesound qualityVSAvoidlatency
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies segmentation by dividing the audio signal into frequency sub-bands and processing them in parallel with different computational parameters. This allows the system to maintain high sound quality for critical frequency ranges while using more efficient, lower-latency processing for other bands, thereby reducing overall system latency.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies full-resolution processing only to specific sub-bands where it is most beneficial, rather than uniformly across all frequencies. This partial application of high-resolution processing maintains sound quality where needed while reducing computational load and latency for the overall system.

Inventive Principle:
Principle #16Partial or excessive action

4Adaptability or versatility

If multiple sound channels are processed simultaneously in live productions, then sound quality coverage is improved, but cost per channel increases due to expensive DSP processors

Engineering Contradiction:
Improvesound quality coverageVSAvoidcost per channel
Core Design Contradiction:
Adaptability or versatilityVSEase of manufacture

Solution Approach 1:

The patent enables efficient multi-channel processing by segmenting each channel's frequency spectrum and processing sub-bands independently at reduced sampling rates. This approach allows multiple channels to be processed simultaneously with fewer computational resources per channel, reducing the need for expensive high-performance DSP processors while maintaining comprehensive sound quality coverage across all channels.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP4380049A1A signal processing method
Publication Date: 2024.06.05 OEKSOUND OY
  • EP4380049A1 patent drawingFigure 1
  • EP4380049A1 patent drawingFigure 2
  • EP4380049A1 patent drawingFigure 3

AI summary

This invention is related to signal processing of music. The invention specifically concerns removal of unwanted artifacts from a music signal. The invention provides a method for attenuating resonances in a sound signal, which method uses subsampling of at least one sub-band of the original signal to save computing resources. As a coarse overview description, the method comprises at least steps in which an incoming signal block is filtered (120) into at least two different frequency sub-band signals, at least one of the sub-band signals is subsampled (121), a resonance analysis step is performed (125) for each sub-band signal, said step resulting in a filter response, a sub-band signal from each sub-band is convolved (130) with the corresponding filter response to produce a filtered sub-band signal, for each subsampled sub-band, the filtered sub-band signal is upsampled (140), and finally the upsampled sub-band signals of subsampled sub-bands and filtered sub-band signals of any non-subsampled sub-bands are combined (150) to form a result signal block. This method saves computing resources by performing analysis based on subsampled sub-band blocks.