Codebook Kalman Filtering for Speech Enhancement

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional single-channel speech enhancement algorithms fail to improve speech intelligibility in the presence of non-stationary background noise, such as babble noise, which is common among hearing aid users.

Innovation Solution

A hearing device and method that utilize a codebook-based approach to determine parameters for Kalman filtering of input signals, enhancing speech intelligibility by separating speech and noise signals using a Kalman filter, particularly in the presence of non-stationary background noise.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional single-channel speech enhancement algorithms are used, then speech quality is improved, but speech intelligibility is not improved in presence of non-stationary background noise

Engineering Contradiction:
Improvespeech qualityVSAvoidspeech intelligibility
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent changes the parameters used in speech enhancement from conventional spectral subtraction parameters to codebook-based parameters that model speech and noise characteristics. By using codebooks containing speech and noise vectors, the system adapts parameters dynamically to match the actual signal characteristics, enabling effective enhancement of speech intelligibility in non-stationary noise environments while maintaining speech quality

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent introduces dynamic adaptation mechanisms where codebooks are updated continuously based on incoming signal characteristics. The speech and noise codebooks are adapted in real-time to track changing noise conditions and speech patterns, allowing the enhancement algorithm to remain effective under non-stationary conditions rather than relying on fixed parameters

Inventive Principle:
Principle #15Dynamics

2Reliability

If codebook based approach processing is performed to determine parameters for Kalman filtering, then speech intelligibility is enhanced, but device complexity increases

Engineering Contradiction:
Improvespeech intelligibilityVSAvoidprocessing complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent uses codebooks that contain pre-stored speech and noise vectors as simplified representations of complex signal characteristics. Instead of performing complex real-time analysis of speech and noise components, the system copies and matches incoming signals against these pre-computed codebook vectors, significantly reducing processing complexity while maintaining accurate parameter estimation for Kalman filtering

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent performs preliminary processing by pre-computing and storing speech and noise codebooks before actual speech enhancement is needed. These codebooks contain pre-analyzed speech and noise characteristics that can be directly used during enhancement operations, eliminating the need for complex real-time decomposition and parameter extraction

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentUS11082780B2Kalman filtering based speech enhancement using a codebook based approach
Publication Date: 2021.08.03 GN HEARING AS
  • US11082780B2 patent drawing
  • US11082780B2 patent drawing
  • US11082780B2 patent drawing

AI summary

A hearing device for enhancing speech intelligibility, the hearing device includes: an input transducer for providing an input signal comprising a speech signal and a noise signal; a processing unit; an acoustic output transducer coupled to the processing unit, the acoustic output transducer configured to provide an audio output signal based on an output signal form the processing unit; wherein the processing unit is configured to determine one or more parameters of the input signal based on a codebook based approach (CBA) processing; and wherein the processing unit is configured to perform a Kalman filtering of the input signal based on the determined one or more parameters so that the output signal has an enhanced speech intelligibility.