Digital companding without range bits involves comparison of a digital input with predetermined boundaries separating a plurality of numerical ranges, detection of a numerical range to which the digital input belongs, generation of an off-set unique to each numerical range, bit-shifting of the digital input, and addition of the bit-shifted input with the generated off-set to provide a companded digital output which is in one-to-one correspondence to the digital input. The off-set values are stored in a memory.