Model-Based Reverberation Equalizer With Temporal Constraint
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing speech recognition systems face challenges in suppressing acoustic signal reverberation, leading to distortion of the desired speech component, especially in reverberant environments, and postfilters often fail to adequately control reverberation artifacts.
Innovation Solution
A method and system that utilize a model-based reverberation equalizer with a temporal constraint to process audio signals, generating outputs that adjust the direct-to-noise ratio, thereby limiting direct sound distortion and enhancing spatial filtering capabilities of postfilters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If reverberation suppression is increased, then the quality of speech recognition improves, but the direct sound component becomes distorted
Solution Approach 1:
The patent applies local quality by differentiating between direct sound and reverberation components, applying different processing strategies to each. The system estimates the direct sound component separately from the reverberation component, allowing selective suppression of reverberation while preserving direct sound integrity. This is achieved through spatial filtering and temporal constraints that target specific acoustic characteristics of each component.
Solution Approach 2:
The patent implements preliminary action by estimating and separating the direct sound component before applying reverberation suppression. The system performs preliminary signal processing to identify and isolate the direct sound path, then uses this information to guide the reverberation suppression process, ensuring that suppression is applied only to the reverberation component rather than the entire signal.
2Object-affected harmful factors
If postfilter is optimized for noise suppression, then noise reduction improves, but reverberation suppression becomes insufficient
Solution Approach 1:
The patent applies segmentation by dividing the acoustic signal into distinct components: direct sound, reverberation, and noise. The system processes each component separately using appropriate algorithms, allowing the postfilter to be optimized for noise suppression while additional processing handles reverberation suppression. This segmentation enables targeted control over which harmful factors are addressed by which processing stages.
Solution Approach 2:
The patent introduces an intermediary processing stage between the postfilter and the final output. This intermediary component, which includes reverberation equalization and temporal constraints, mediates between the noise-suppression optimized postfilter and the final speech signal, correcting the insufficient reverberation suppression while preserving the noise reduction benefits.
3Object-generated harmful factors
If spatial filtering is applied to suppress reverberation, then reverberation suppression improves, but direct sound distortion increases
Solution Approach 1:
The patent applies parameter changes by dynamically adjusting filtering parameters based on the estimated direct sound component. The system modifies the spatial filtering characteristics in real-time, adapting the filter coefficients and constraints based on the current signal conditions and the estimated direct sound path. This allows optimization of reverberation suppression parameters while maintaining direct sound quality through adaptive control.
Data Source
AI summary
A method, computer program product, and computer system for addressing acoustic signal reverberation is provided. Embodiments may include receiving, at one or more microphones, a first audio signal and a reverberation audio signal. Embodiments may further include processing at least one of the first audio signal and the reverberation audio signal. Embodiments may also include limiting a model based reverberation equalizer using a temporal constraint for direct sound distortions, the model based reverberation equalizer configured to generate one or more outputs, based upon, at least in part, at least one of the first audio signal and the reverberation audio signal.