A packet loss concealment method base on GWSOLA algorithm and signification transient detectd

We present a new packet loss concealment method based on time-scale modification (TSM) of voice signal. We introduce the gain control into the WSOLA technique and we called it GWSOLA. GWSOLA method can reconstruct the voice signal of the lost frames. We also present two detection methods that can classify the whether lost frame is location at the voice signification transmit (VST) section or not, and then use different gain to reconstruct the lost frame. By applying these techniques, we can reduce quality degradation of the decoded speech. Mean opinion score (MOS) show that these proposed methods superior to the traditional WSOLA algorithm.