CELP based speech coding with fine granularity scalability

General audio and video with fine granularity scalability (FGS) has become favored in next generation multimedia coding standards due to its high flexibility in channel rate adaption. However, the FGS phenomenon has not yet been fitted into existing speech codecs. We introduce the FGS feature to the code excited linear prediction (CELP) based speech coding algorithm by adjusting the amount of transmitted fixed excitation information. We further improve the algorithm by relaxing the constraints and re-ordering the sequence of pulses. To achieve this target, we need to make modifications to the conventional coding algorithm, but the computation overhead is little and affected modules are few. As a consequence, developers can, in a short time, easily migrate their existing codec to one with the FGS advantage.