ADAPTIVE DIGITAL DELTA MODULATION FOR VOICE TRANSMISSION

摘要

TITLE ADAPTIVE DIGITAL DELTA MODULATION FOR VOICE TRANSMISSION INVENTORS David E. Dodds Andrzej M. Sendyk Donald B. Wohlberg The adaptive delta modulation system represents an approximation of an analog voice signal by a stream of signal binary bits each occupying a fixed time period. This stream of binary signals is produced by periodically sampling the analog signal and generating a binary bit for each period, the logic level of the binary bit being dependent on whether the sampled signal is greater or smaller than the approximate signal of the previous sample. The decoder apparatus which converts the stream of binary bits back to the approximate analog signal includes a capacitor integrator which is periodically charged or discharged by a pre-determined amount in response to each bit so that the signal on the capacitor follows approximately the value of the original analog signal. The charging or discharge during each period is determined by the logic level of the binary bit. In addition, the step size increases or decreases approximately exponentially for successive periods with the step increasing in response to successive similar signal binary bits and decreasing in response to successive dissimilar bits. This increase or decrease in step size is achieved by storing a binary step size number S in a register and adding or subtracting a fraction of the number to or from itself during the particular period, this produced a new step size number for each period. In addition, each added fraction may be increased by a fixed least significant number to enhance step size recovery between a transmitter and a receiver. The charging and discharging of the capacitor integrator is accomplished by a pulse that has a width which is directly related to the value of the S number. The width of this control pulse controls the conduction time of constant current sources which are connected to the integrating circuit to charge or discharge the integrating capacitor and produce the approximate signal.