Application of pulse-code modulation to an integrated telephone network. II. Transmission and encoding

For Part I see abstract 1963B13777. The conversion from speech to pulse-code modulated multiplex signals, which is performed in the concentrator, uses the intermediary of pulse-amplitude modulation. Sampling is done at 8000-10000 c/s which is higher than the required double upper speech frequency but leads to less expensive filters. Conversion to pulse-code signals is done in the coder which includes a compression stage which makes it possible to reduce the number of quantization steps. Compression characteristic consists of five straight line segments subtending a logarithmic characteristic; the central segment has 64 quantization steps and corresponds to 25dB improvement in the signal-to-quantization noise factor for weak signals; other segments have 16 steps each. Accumulation in both coder and decoder is in a digital memory.