The forward-backward recovery sub-codec (FB-RSC) method: a robust form of packet-loss concealment for use in broadband IP networks

The FB-RSC method of speech coding provides improved speech quality under packet-loss conditions. Each packet has a maximum of three codes, a main code, a forward sub-code, and a backward sub-code. The main code represents the current frame. The forward and backward sub-codes represent the next and previous frames. The necessity of the sub-codecs is determined by a sub-codec selector, which considers the SNR of the original and synthesized signals in this determination. A relatively low-compression and high-quality form of coding is used in the main codec, while coding with greater compression is used in the sub-codecs. We examined the quality of the proposed method for random loss of individual packets and pairs of consecutive packets. The result shows that our method significantly improves the concealment of packet loss.