Two-Stage Noise Estimation for Hearing Aids
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing noise reduction methods for hearing devices face challenges in accurately estimating noise power, particularly during speech activity, leading to unsatisfactory noise suppression and the introduction of artifacts like 'musical tones', as they struggle to balance slow adaptation during speech and fast adaptation when no speech is present.
Innovation Solution
A two-stage estimation method is employed, where a first estimation algorithm provides parameters for a second estimation algorithm to adapt noise reduction based on the input signal's characteristics, allowing for situation-specific noise estimation and reduction, potentially using the same or different algorithms in a time division multiplexing mode to enhance noise power estimation and reduce artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single noise estimation algorithm is used, then the device complexity is reduced, but the noise suppression quality deteriorates due to inability to adapt to different acoustic situations
Solution Approach 1:
The noise estimation process is divided into two separate algorithms: a first noise estimation algorithm that operates continuously and a second noise estimation algorithm that is activated under specific conditions (when speech activity is detected). This segmentation allows each algorithm to be optimized for its specific function, improving overall estimation accuracy without requiring a single overly complex algorithm to handle all scenarios.
Solution Approach 2:
The system dynamically switches between different estimation algorithms based on the detected acoustic situation. When speech activity is detected, the system transitions from the first estimation algorithm to the second estimation algorithm, which is parameterized using information from the first algorithm. This dynamic adaptation allows the system to maintain high estimation accuracy across varying acoustic conditions while keeping individual algorithm complexities manageable.
2Speed
If noise estimation adapts quickly, then noise suppression responsiveness is improved, but speech quality deteriorates due to misclassification of speech components as noise
Solution Approach 1:
The adaptation speed is dynamically controlled based on the acoustic situation. The first estimation algorithm provides a baseline that adapts quickly to changing noise conditions. When speech activity is detected, the system switches to the second estimation algorithm which uses parameters from the first algorithm to maintain appropriate adaptation speed while preventing misclassification of speech components. This dynamic control resolves the contradiction between fast adaptation and speech preservation.
Solution Approach 2:
The second estimation algorithm is parameterized using information from the first estimation algorithm, creating a feedback mechanism. The first algorithm's output serves as input parameters for the second algorithm, allowing the system to leverage the fast adaptation capabilities of the first algorithm while using the second algorithm's speech-aware processing to prevent misclassification. This feedback loop ensures both fast responsiveness and reliable speech preservation.
3Reliability
If noise estimation is conservative (slow adaptation), then speech quality is maintained, but noise suppression effectiveness deteriorates during non-speech periods
Solution Approach 1:
The estimation process is segmented into two algorithms with different adaptation characteristics. The first algorithm provides conservative, speech-preserving estimation during speech activity, while the second algorithm (activated during non-speech periods) can adapt more aggressively to track noise changes. This segmentation allows the system to maintain speech quality during speech periods while achieving effective noise suppression during non-speech periods, resolving the productivity-reliability contradiction.
4Measurement precision
If multiple estimation algorithms are used, then noise suppression quality is improved through situation-specific estimation, but device complexity increases
Solution Approach 1:
The system uses two distinct estimation algorithms segmented by their activation conditions. The first algorithm operates continuously providing baseline estimation, while the second algorithm is activated only when speech activity is detected. This segmentation allows the system to achieve situation-specific estimation accuracy without requiring all algorithms to run simultaneously, thereby managing computational complexity while improving measurement precision.
Solution Approach 2:
The first estimation algorithm performs preliminary noise estimation that provides parameter information for the second algorithm. This preliminary action allows the second algorithm to be efficiently parameterized without requiring independent training or complex initialization, reducing the overall computational burden while maintaining the benefits of multiple algorithms for different situations.
Data Source
AI summary
The noise reduction for signals which can contain speech at least part of the time is to be improved. To this end a hearing apparatus and especially a hearing device with a first estimation device for estimating a first value of an input signal with a first estimation algorithm and a noise reduction device for reducing noise in the input signal are provided. A second estimation device, which is parameterized with the estimated first value, is used for estimating a second value of the input signal with a second estimation algorithm. The noise reduction device receives the estimated second value from the second estimation device for reducing the noise. The two-stage estimation method enables an adaptive estimation to be carried out which is always currently adapted to an input signal.