2 results about "PESQ" patented technology

In the low signal-to-noise ratio condition, aiming at the problems that the traditional neural network speech feature extraction is insufficient, and the speech enhancement effect needs to be improved, based on empirical mode decomposition (EMD), temporal convolution network (TCN) and gated convolution recurrent neural network (GCRN), and combining with feature fusion module (FFM), the application proposes a speech enhancement model of adaptive mean median empirical mode decomposition-multilayer gated feature fusion module convolutional recurrent neural network (ME-MGFCRN). The network model adopts the frequency learning strategy to learn the low frequency feature and the high frequency feature, that is, the TCN and the MGFCRN network are used to obtain the low frequency and the high frequency feature, and the two groups of features are processed through the FMM, so as to realize the speech enhancement in the feature mapping mode. The model proposed in the application carries out the ablation experiment and the comparison experiment on the data set, and uses the PESQ, fwSegSNR and STOI indexes to evaluate the speech enhancement effect. Research shows that under different noise environments and different signal-to-noise ratios, the model proposed in the application is improved compared with other baseline models, especially under the low signal-to-noise ratio condition of SNR of-5dB, the fwSegSNR and PESQ are improved by more than 0.86dB and 0.02 respectively compared with other baseline models.

This invention provides a deep echo cancellation method based on cross-domain prior interaction gating and feature decoupling, comprising: preprocessing the acquired near-end microphone signal and far-end reference signal to obtain a complex spectrum with uniform time-frequency resolution; constructing an echo cancellation model based on the complex spectrum using a cross-modal gating attention mechanism, a conditional Transformer, and a multi-branch decoder; and performing echo cancellation on the newly acquired mixed speech signal based on the echo cancellation model to obtain enhanced near-end speech. This method can still learn echo saliency maps through pseudo-labels / weak supervision in unlabeled scenarios, and inference only requires a single forward computation, thereby improving PESQ / STOI and ERLE and reducing speech distortion.