In-Ear Speech Detection for Adaptive Noise Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audio wearable devices struggle with effective noise reduction in dynamic environments due to resource constraints and limitations in power and processing capabilities, leading to diminished audio quality and user frustration.
Innovation Solution
Integration of deep neural networks (DNNs) into in-ear devices for real-time noise reduction, utilizing DNNs to distinguish between speech and noise and adjust noise reduction aggressiveness based on speech presence probability (SPP), with adaptable and dynamic noise reduction capabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If deep neural networks are integrated into in-ear devices for real-time noise reduction, then audio quality is improved, but device complexity increases
Solution Approach 1:
The system segments the noise reduction task by using a DNN to first classify audio segments as speech or noise, then applying appropriate processing only to noise segments. This segmentation allows the complex DNN to work with simplified, targeted processing rather than attempting to process all audio uniformly.
Solution Approach 2:
The noise reduction aggressiveness is dynamically adjusted based on the DNN's real-time classification output. When speech is detected, the system reduces noise reduction strength to preserve speech quality. When noise is detected, the system increases noise reduction strength. This dynamic adaptation allows the system to achieve high audio quality without requiring continuously maximum processing power.
2Object-affected harmful factors
If noise reduction aggressiveness is increased to suppress unwanted noise, then noise suppression is improved, but speech quality deteriorates
Solution Approach 1:
The DNN continuously monitors the audio stream and provides feedback about whether the current segment contains speech or noise. This feedback loop allows the noise reduction system to adjust its aggressiveness in real-time, increasing it when noise is detected and decreasing it when speech is detected, thereby preventing speech quality deterioration.
Solution Approach 2:
The system changes the noise reduction parameter (aggressiveness level) based on the DNN's classification output. By dynamically adjusting this parameter between high and low states depending on whether speech or noise is detected, the system achieves effective noise suppression without compromising speech quality.
3Productivity
If deep neural networks are used for real-time speech and noise distinction, then noise reduction effectiveness is improved, but processing power requirements increase
Solution Approach 1:
The system applies the computationally intensive DNN classification only to portions of the audio signal (specifically, only when needed based on detection thresholds), rather than continuously processing all audio at maximum complexity. This partial application reduces average processing power requirements while maintaining high noise reduction effectiveness when required.
Solution Approach 2:
The processing power consumption is dynamically adjusted based on the detected audio conditions. When the DNN detects noise that requires suppression, the system increases processing power. When speech is detected or the audio is quiet, the system reduces processing power. This dynamic scaling allows the system to achieve high productivity when needed while conserving resources during normal operation.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
An ear-wearable device includes at least one microphone, a receiver that is placed within an ear of a user. An audio processing path of the device receives an audio signal from the at least one microphone and reproduces the audio signal at the receiver. The ear-wearable device includes a deep neural network (DNN) that is coupled to the audio processing path and is trained to distinguish between speech and noise in the audio signal. A speech presence probability (SPP) is determined based on an output of the DNN. The ear-wearable device includes a noise reduction system coupled to the audio processing path and is operable to perform noise reduction on the audio signal. The noise reduction system is coupled to receive the SPP from the DNN and change an aggressiveness of the noise reduction based on a value of the SPP.