Companding
In communications engineering, companding is a process that can be used to mitigate the adverse effects of a limited dynamic range in a transmission or storage process for audio signals. This improves the signal-to-noise ratio and prevents overdriving. The expression is a suitcase word made up of " compression " and " expansion ". The procedure is u. a. used on the phone .
Demarcation
Companding is a non-linear , static process in which the signal voltage is pre-distorted ("compressed") by a specified characteristic and then transmitted to the target. In this form the signal is very heavily distorted. During playback, the signal goes through an "expander" with a complementary characteristic, where it is converted back into the original curve shape. The loss of quality is usually small.
Companding is often confused with compression , in which a control voltage generated by an envelope demodulator changes the gain of a (linearly operating) Voltage Controlled Amplifier relatively slowly (in the Hertz range) so that the output voltage remains within a desired range. This can either avoid overmodulation or (with the Dolby system) reduce disturbing noise. The sometimes clearly audible change in the original is accepted here with approval. The amplitude changes caused by a compressor can only be partially reversed by an expander .
Working principle
The very large dynamic range of speech is pre-distorted in telephone systems using the A-law method. Since the characteristic curve is difficult to generate with analog means, it is only made after digitization. The analog-digital converter initially works with 12-bit accuracy, which corresponds to a dynamic range of 72 dB. The A-law method reduces the amplitude range to ± 42 dB, which is why the data can be transmitted with 8-bit accuracy. At the destination, the data is first converted to the original 12-bit accuracy and then converted back into the voice signal by a digital-to-analog converter .
The method described is referred to as predistortion of amplitudes in the time domain . To reduce high-frequency noise, pre-emphasis is used, which can be described as pre-distortion of amplitudes in the frequency domain .