Buffered Speech Recognition Noise Suppression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Speech recognition systems face challenges in noisy environments and during transitional periods when users change their position relative to devices, leading to suboptimal processing and potential distortion of speech signals due to the time required for algorithms to adjust parameters.
Innovation Solution
A system and method that involves buffering sensor data, including acoustic signals, to produce new parameters for processing, which enables faster-than-real-time noise suppression and direction-of-arrival estimation, allowing for improved separation of clean voice from noise and adaptation of processing parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If pre-processing algorithms use multiple microphones and adjust parameters in real-time, then the system can adapt to changing voice directions, but the processing is not optimal during transitional periods and may degrade the speech signal
Solution Approach 1:
The system performs preliminary parameter adjustment during a pre-processing stage before the main speech recognition processing. This allows the algorithm to adapt to new voice directions and optimize parameters in advance, so that when actual speech recognition occurs, the parameters are already optimized and no degradation occurs during transitional periods.
2Measurement precision
If the processor adjusts internal parameters to account for new orientation, then optimal processing is achieved, but the beginning of the utterance can be distorted until the processor settles on optimal parameters
Solution Approach 1:
The system performs parameter optimization in advance during a pre-processing stage before the actual utterance is processed. This preliminary action ensures that when speech recognition begins, the parameters are already optimized, eliminating the distortion period that would otherwise occur at the beginning of the utterance.
3Speed
If speech recognition processes audio in real-time, then response speed is maintained, but noise suppression and direction-of-arrival estimation cannot be performed optimally
Solution Approach 1:
The system segments the speech processing into two distinct stages: a pre-processing stage for offline optimization of parameters including noise suppression and direction-of-arrival estimation, and a main processing stage for real-time speech recognition. This segmentation allows computationally intensive operations to be performed in advance without impacting real-time response requirements.
Solution Approach 2:
The system performs noise suppression and direction-of-arrival estimation in advance during the pre-processing stage, before real-time speech recognition begins. This preliminary computation of acoustic parameters enables optimal noise suppression accuracy to be achieved without compromising the speed requirements of real-time processing.
Data Source
AI summary
Systems and methods for assisting automatic speech recognition (ASR) are provided. An example system includes a buffer operable to store sensor data. The sensor data includes an acoustic signal, the acoustic signal representing at least one captured sound. The system includes a processor communicatively coupled to the buffer and being operable to store received sensor data in the buffer. The received sensor data is analyzed to produce new parameters associated with the sensor data. The buffered sensor data is processed based at least on the new parameters. The processing may include separating clean voice from noise in the acoustic signal. The processor is further operable to provide at least the processed sensor data (for example, the clean voice) to an ASR system operable to receive and process the processed sensor data at a speed faster than real time. The new parameters may also be provided to the ASR system.