摘要 |
A modulated signal carrying speech or other digitally coded information is demodulated (1): a likelihood signal l(t) is produced (e.g. in Viterbi decoder (4)) for each demodulated symbol, indicating its reliability. This is used to update probabilities Pg(t), Pb(t) of the transmission channel being in a relatively error-free or error-prone state, using a Viterbi algorithm, in dependence on stored parameters representing estimates of the probability transitions between the states. A decision signal E(t) is generated dependent on the relative magnitudes of the probabilities Pg(t), Pb(t). The result may be used to inhibit the use of the affected bits, e.g. by supplying a speech decoder (9) with parameters (stored in 82a etc.) from an earlier speech frame. |