Causal TCN for Multichannel Speech Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing speech detection methods, particularly in noisy and reverberant environments, face challenges in distinguishing speech from non-speech due to the complexity and training requirements of recurrent neural networks (RNNs), which can lead to inefficiencies and inaccuracies in real-time processing.
Innovation Solution
The implementation of a causal temporal convolutional network (TCN) for real-time or near real-time speech detection, utilizing dilated convolutions to reduce complexity and memory requirements, allowing for faster training and more accurate classification of speech segments within multichannel audio inputs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If recurrent neural networks (RNNs) are used for speech detection in noisy environments, then speech detection accuracy is improved, but device complexity and training time increase significantly
Solution Approach 1:
The patent changes the fundamental parameters of the neural network architecture by replacing RNNs with Temporal Convolutional Networks (TCNs). This involves changing the network structure from recurrent to convolutional, modifying the computational approach from sequential to parallel, and adjusting the parameter update mechanism to eliminate the vanishing gradient problem while maintaining speech detection accuracy in noisy environments
2Measurement precision
If recurrent neural networks (RNNs) are used for speech detection, then speech detection accuracy is improved, but training time and memory requirements increase
Solution Approach 1:
The patent substitutes the mechanical recurrent structure of RNNs with a convolutional architecture that uses dilated convolutions and causal masking. This replacement enables parallel computation during training, eliminating the sequential processing bottleneck of RNNs and significantly reducing training time while preserving the ability to capture temporal dependencies for accurate speech detection
3Productivity
If simple energy detector is used in low-noise environments, then processing speed is improved, but speech detection accuracy deteriorates in noisy environments
Solution Approach 1:
The patent implements a dynamic approach by using Temporal Convolutional Networks that can adapt to varying noise conditions. The TCN architecture dynamically processes temporal patterns in the audio signal, maintaining high processing speed through efficient convolution operations while achieving accurate speech detection in both low-noise and noisy environments by learning robust temporal features
Data Source
AI summary
Embodiments of the disclosure provide systems and methods for speech detection. The method may include receiving a multichannel audio input that includes a set of audio signals from a set of audio channels in an audio detection array. The method may further include processing the multichannel audio input using a neural network classifier to generate a series of classification results in a series of time windows for the multichannel audio input. The neural network classifier includes a causal temporal convolutional network (TCN) configured to determine a classification result for each time window based on portions of the multichannel audio input in the corresponding time window and one or more time windows before the corresponding time window. The method may additionally include determining whether the multichannel audio input includes one or more speech segments in the series of time windows based on the series of classification results.