Speech Quality Detection for Adaptive Microphone Audio Processing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing audio devices fail to effectively detect and improve audio quality in real-time, particularly due to factors like background noise, reverberation, and interfering speech, leading to degraded speech quality and discomfort for users.

Innovation Solution

An audio device equipped with a non-intrusive quality detection model, utilizing a neural network to assess speech quality without a reference signal, and dynamically adjusts processing to enhance audio quality through noise suppression and echo cancellation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Area of stationary object

If the user is located far from the microphone, then the coverage area is improved, but the speech quality degrades due to low signal to noise ratio and reverberation

Engineering Contradiction:
Improvecoverage areaVSAvoidspeech quality
Core Design Contradiction:
Area of stationary objectVSMeasurement precision

Solution Approach 1:

The system uses a non-intrusive quality detection model to continuously monitor speech quality parameters (SNR, reverberation, interfering speech) and provides feedback to control the activation of digital signal processing features. This closed-loop feedback mechanism enables dynamic adaptation to maintain speech quality across varying distances and acoustic environments.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The system dynamically adjusts the processing of microphone input signals based on real-time quality parameters. Digital signal processing features such as noise suppression, echo cancellation, and interferer suppression are selectively activated or deactivated according to the detected speech quality, allowing the system to optimize performance for both near-field and far-field scenarios.

Inventive Principle:
Principle #15Dynamics

2Measurement precision

If digital signal processing is applied to improve speech quality, then the speech quality is improved, but the device complexity increases

Engineering Contradiction:
Improvespeech qualityVSAvoidprocessing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

Instead of applying digital signal processing continuously, the system selectively activates processing features only when quality parameters indicate degradation. The non-intrusive quality detection model identifies specific conditions (low SNR, high reverberation, interfering speech presence) and triggers appropriate processing interventions, avoiding unnecessary computation when speech quality is already acceptable.

Inventive Principle:
Principle #16Partial or excessive action

Solution Approach 2:

The system changes processing parameters dynamically based on detected quality metrics. The non-intrusive quality detection model monitors parameters such as SNR, reverberation time, and interferer levels, and adjusts the activation state of digital signal processing features accordingly, transforming the system from a static to a dynamically adaptive processing architecture.

Inventive Principle:
Principle #35Parameter changes

3Measurement precision

If multiple digital signal processing features are activated to suppress noise and echo, then the speech quality is improved, but the loss of information increases due to over-processing

Engineering Contradiction:
Improvespeech qualityVSAvoidaudio information loss
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The non-intrusive quality detection model provides continuous feedback on speech quality parameters to control the activation of digital signal processing features. This feedback mechanism ensures that processing is applied only when and where needed, preventing over-processing that would degrade audio information while still improving speech quality in degraded conditions.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The system applies digital signal processing selectively rather than universally. By using quality parameters to determine when noise suppression, echo cancellation, or interferer suppression should be activated, the system avoids excessive processing that would remove legitimate audio information while still providing enhancement where quality degradation is detected.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP4207194B1Audio device with audio quality detection and related methods
Publication Date: 2026.04.29 GN AUDIO AS
  • EP4207194B1 patent drawingFigure 1
  • EP4207194B1 patent drawingFigure 2
  • EP4207194B1 patent drawingFigure 3

AI summary

An audio device and related methods for speech quality detection are disclosed, the audio device comprising an interface, a processor, a memory and one or more microphones, wherein the audio device is configured to obtain, via the interface, a microphone input signal from one or more microphones including a first microphone; process the microphone input signal for provision of an output signal; determine, using a non-intrusive quality detection model, one or more quality parameters including a first quality parameter indicative of a speech quality associated with the output signal; control processing of the microphone input signal based on the first quality parameter; and transmit, via the interface, the output signal.