Use of low-delay code-excited linear prediction technology in circuit multiplexed networks

The LD-CELP (code excited linear prediction) algorithm was adopted by the CCITT, as a Recommendation G.728 for the coding of speech with toll quality at 16 kbit/s. The operation of the LD-CELP algorithm at 12.8 kbit/s is described, and its performance is assessed both with voice and nonvoice signals in single and interconnected network configurations. The 12.8 kbit/s LD-CELP codec is found to perform equivalently to 24 kbit/s ADPCM (adaptive differential pulse code modulation). It is also shown that 12.8 kbit/s LD-CELP is acceptably transparent to network signaling. As a result, it can be concluded that the operation of the LD-CELP algorithm at 12.8 kbit/s presents a viable option for inclusion within 16-kbit/s-based DCME (digital circuit multiplication equipment) or PCME (packet circuit multiplication equipment) overload strategies.<>