Doing some more experimenting. Decided to implement windowing in AGC and multiband 1 (as used in some Orban and Omnia processors.) More apparent dynamic range and punch by allowing more peaks to pass through to the the final multi-band segment and clipper(or in my case, the limiter--no final clipper in use.) I am processing older tunes--before mastering of songs clipped, squashed and destroyed the music. As always with this intense amount of processing the cleaner the input audio the more processing can be used with a cleaner output. I am using WAV files ripped from my personal collection. The indicated level on the output hovers around -7 ITU.
Edit:The ear has a built-in AGC--it operates with essentially an RMS response. The brain doesn't "turn down the gain" on short transient peaks, therefore the transients are perceived in the brain as "louder." In the FM broadcast scenario, it is OK to let those transients hit the clipper a bit harder(composite or other hard clipper) because they are so short they don't add much audible distortion yet they can allow a higher legal level of modulation.
|