Posture-Aware Voice Processing With Adaptive Correction Filters

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing audio signal processing systems fail to account for the posture of a talker, leading to inadequate voice enhancement and quality.

Innovation Solution

An audio signal processing method that estimates posture information of a talker from an image, generates a correction filter based on this information, and applies filter processing to enhance voice quality.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If audio signal processing is performed without considering talker posture, then the processing system is simple, but voice quality and enhancement accuracy deteriorate

Engineering Contradiction:
Improvevoice qualityVSAvoidprocessing system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent introduces an image processing module as an intermediary that captures and analyzes the talker's posture from video feed. This separate module processes visual information to extract posture parameters, which then serve as inputs for adjusting audio processing parameters. This intermediary approach enables accurate voice quality enhancement while maintaining modular system architecture.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The system dynamically changes audio processing parameters based on detected posture parameters. When the talker's posture changes (e.g., head orientation, body position), the system adjusts microphone selection, gain levels, and beamforming parameters accordingly. This parameter adaptation allows the system to maintain high voice quality without requiring a completely complex reconfiguration.

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If posture information is obtained through image processing, then voice enhancement accuracy improves, but processing time and computational load increase

Engineering Contradiction:
Improvevoice enhancement accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system performs preliminary posture detection and analysis in parallel with audio signal acquisition. By pre-processing the video feed to extract posture information before audio enhancement is applied, the system avoids sequential processing delays. The posture parameters are prepared in advance and ready for immediate use when audio processing needs adjustment.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The image processing module focuses on detecting only the critical posture parameters needed for audio enhancement (such as head orientation and body position) rather than performing complete facial or body analysis. This selective detection approach provides sufficient accuracy for voice enhancement while significantly reducing computational load and processing time.

Inventive Principle:
Principle #16Partial or excessive action

3Reliability

If correction filters are generated based on posture information, then voice quality compensation improves, but system complexity increases

Engineering Contradiction:
Improvevoice quality compensationVSAvoidfilter generation complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The system generates correction filters by adjusting parameters such as gain levels, frequency response characteristics, and beamforming weights based on posture parameters. Instead of creating entirely new complex filter structures, the system modifies existing filter parameters in response to posture changes, simplifying the filter generation process while maintaining effective voice quality compensation.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The system implements a feedback mechanism where posture detection results continuously inform correction filter adjustments. The detected posture parameters feed into the filter generation module, which automatically adapts filter characteristics to compensate for voice attenuation and reverberation caused by the talker's current posture. This closed-loop approach improves reliability without requiring manual intervention or complex decision logic.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP3982363B1Audio signal processing method and audio signal processing apparatus
Publication Date: 2025.09.03 YAMAHA CORP
  • EP3982363B1 patent drawingFigure 1
  • EP3982363B1 patent drawingFigure 2
  • EP3982363B1 patent drawingFigure 3

AI summary

An audio signal processing method includes receiving an audio signal corresponding to a voice of a talker (S11), obtaining an image of the talker (S12), estimating posture information of the talker using the image of the talker (S23), generating a correction filter according to the estimated posture information (S14), performing filter processing on the audio signal using the generated correction filter (S15), and outputting the audio signal on which the filter processing has been performed (S16).