Verfahren zur Sprachkodierung mittels linearer Prädiktion

摘要

The method involves determining a spectral state (Y) from first and second states with proportionately less low frequency energy contained in the first state than the second. One of two quantisation modes is applied to determine quantisation values of the coefficients of a synthesis filter. The determined spectral state is not modified when the energy is less than a certain threshold. Each frame is detected to determine whether the signal is in the first or second state and the state is modified if several successive frames show a different signal condition to the previous state as registered by a counter.