Error detection and error concealment for encoded speech data

An improved error detection and error concealment for Viterbi decoding of convolutionally encoded data is provided. The most sensitive part of the data is parity encoded and sent with parity and this data with the next most sensitive data are convolutionally encoded and sent with the least sensitive data over a transmission channel to a receiver. At the receiver the convolutionally encoded data is decoded using the Viterbi algorithm. The decoder compares the parity computed from decoded data with the decoded parity and if they are not equal generates a Bad Frame Indicator (BFI) flag and also determines which decoded parameters are likely bad and hence generates a Bad Parameter Indicator (BPI) flag for those parameters, by determining the confidence levels for the parameters and comparing against pre-selected thresholds. The decision to discard a decoded parameter is dependent on the BFI flag and the BPI flag of that parameter.