A Temporal-Spectral Generative Adversarial Network based End-to-end Packet Loss Concealment for Wideband Speech Transmission

This project is the implementation of "A Temporal-Spectral Generative Adversarial Network based End-to-end Packet Loss Concealment for Wideband Speech Transmission", which is submitted to Journal of the Acoustical Society of America. We also provide some test results of the Voice Bank corpus in "data". (Loss rate ranging from 5% to 30%)

The uploaded code is the original version of the non-causal framework and differs significantly from the causal framework and subsequent versions. And these methods are required not to be made public.