|
 Kyma Forum
 Tips & Techniques
 Dynamic Gain Compression
|
| next newest topic | next oldest topic |
| Author | Topic: Dynamic Gain Compression | |
|
David McClain Member |
posted 19 December 2000 02:32
 
Here is one that behaves as one might expect. The gain at max signal is unity, and increases as the signal mean amplitude diminishes out to the threshold. At threshold there is maximum gain, and below that it acts like a noise gate, in this case simply chopping the signal off. You may want to implement different behavior below threshold. In terms of dB measure, this compressor with a compression ratio R only allows the signal to drop 1 dB for every R dB drop in the mean input signal. No post-compression gain should be necessary. This is dynamic-range compression. Perhaps someone from SSC could explain the reasoning behind their implementation of compression? - DM IP: Logged | |
|
SSC Administrator |
posted 19 December 2000 09:28
The idea is to compress the dynamic range. For example, large peaks should be attenuated, so there is less of a difference between the threshold and maximum amplitudes. In the Kyma compressor, everything below the threshold passes through unaltered. Everything above the threshold is inversely attenuated (the higher the amplitude, the more attenuation). The overall effect is to push the higher amplitudes down so they are closer in value to the threshold amplitude (i.e. a smaller "range" in the dynamics). There is a variable Gain at the output, so the entire signal can be amplified. You can get the overall signal to sound "louder" this way, because you no longer have those peaks that would otherwise have clipped when you amplified the signal that much. IP: Logged | |
|
David McClain Member |
posted 19 December 2000 13:53
What you have described is indeed gain compression, but that is not what the Kyma compressor seems to do... Instead, it appears to mulitply everything above threshold by some fixed value (the ratio parameter) and then adjusts output gain. This is simply a one-level gain switch. Are you actually suggesting that the Kyma compressor does what my patch above does, i.e., gain adjustment in log space (effectively an exponentiation of the gain control)? Have I actually reinvented the Kyma compressor? I would like to think so, so I don't have to work as hard to accomplish this, but somehow I think we have two different notions of compression. Your response is most welcome! (PS: this is one of those areas where it would really help to see either pseudo code or some equations!) - DM IP: Logged | |
|
David McClain Member |
posted 19 December 2000 14:44
As I said before, I have always been confused by the Kyma implementation. Perhaps, once again, it is a mismatch in our respective languages. The answer to the following question would be helpful... If I use a Kyma compressor with a ratio of 5 and a threshold of -40 dB, in order to get the 0 dB inputs to be 0 dB on output, by how much should I adjust compressor post-gain? Is it I think the answer should be (a). Thanks, - DM IP: Logged | |
|
David McClain Member |
posted 19 December 2000 16:13
Well, well, well... In the continuing saga, I put the results on a spectrum analyzer and fed the system with pink noise. Lo and behold, the correct answer to the question above is "(c) none of the above". It turns out that a 0 dB signal injected needs 0 dB post-compressor gain. And apparently, your compressor behaves exactly like I would want. A decrease of N dB gets compressed to a decrease of N/R dB, where R is the compression ratio. I told you there was something confusing about this... - DM IP: Logged | |
|
JohnCowan Member |
posted 19 December 2000 23:54
I've been wondering about designs of various hardware compressors such as the RNC or the new one from TC Electronics. Does anyone know what the conceptual designs are for these units or any others? Of course, the idea would be to impliment something in kyma and maybe add a few new twists. Since I have not done much live recording, I am only familiar with the basics of compressors. John IP: Logged | |
|
SSC Administrator |
posted 20 December 2000 10:46
David -- The Kyma compressor is a classic dynamic range controller in its behavior and implementation. The best way to see this behavior is to apply a known envelope to a steady-state signal and then observe the amplitude envelope of the compressor's output. John -- You can find many examples of twists on compression (including adding filters to the compressor's side chain and multiband compressors, etc.) in the EQ Sound file in the Effects Processing folder of the Kyma Sound library. IP: Logged |
All times are CT (US) | next newest topic | next oldest topic |
  |
|
Contact Us | Symbolic Sound Home
This forum is provided solely for the support and edification of the customers of Symbolic Sound Corporation.
Ultimate Bulletin Board 5.45c