Audio compression usually wants to keep an undistorted signal and primarly modify the signal envelope. A log amp would cause significant signal distortion.It was already answered regarding logaritmic amplifier.
You possibly refer to the µ-law codecs used in telephone systems? In contrast to the DC logarithmic amplifiers quoted by ninju, a logarithmic compression characteristic for an audio signal must be bipolar, like y = sign(x)*log(abs(x)). And it would need an inverse decompression characteristic at the receiver.Logaritmic compression was used at hardware level in some broadcast systems, not in modern storage codecs.
Sorry, to blurry to identify any details.also this block diagram i posted can be corrected
The compressor circuits I know (and used in some applications) are feedback AGC circuits with a transistor utilized either as variable resistor or variable gain amplifier. They produce some distortions but have still almost linear behaviour in the AC signal path.I noticed the LOG compressor used to limit amplitude in a RF AM transmiter circuit much time ago
I don't agree. Review the professional audio dynamic compressors, particularly those described in the link given by rogs. It's true of course, that recent instruments are mostly using DSP techniques which gives some special advantages like applying delay to the signal path. But nervertheless, high performance devices have been designed in a fully analog way for decades, and e.g. used in recording or broadcast studios. Be sure, that no audible distortions are tolerated in these applications.As far as analog goes, i think any compression circuit is meant to distort the signal.
Distortion - Wikipedia, the free encyclopediaDistortion refers to any kind of deformation of a waveform, compared to an input, usually Clipping, harmonic distortion and intermodulation distortion (mixing phenomena) caused by non-linear behavior of electronic components and power supply limitations.