I've always wondered about multi-band look-ahead limiters with "true peak limitation".
For simplicity - let's discuss a mono (single channel) look-ahead limiter with two bands: LF and HF.
The design request of this system is that the output is bi-polar limited between: [ -1.0 .. 1.0], so not even a single sample exceeds 1.00 or ( - 1.00 ).
So, say I had designed a system which limits the LF band and another one which limits the HF band. Now each band is limited within range, but once I sum both outputs, I can still add "1.00' from the HF with "1.00" from the LF and get "2.00" in the summed output.
How can I solve this? Add another limiter on the summed output? Anything more sophisticated ?
Multi channel limiter
-
- KVRist
- 296 posts since 1 Apr, 2009 from Hannover, Germany
Basically you can't solve this. And I would argue that it doesn't even make sense to try. What result or improvement do you expect from something like that?
You could just limit the two bands to [-0.5 .. 0.5] and you'd be sure you never overshoot the final signal. But still it would be unclear if such a thing is even desirable and why.
The thing is, if you split a usual music signal in two bands and limit separately, the biggest contribution to the peak level will be in the low band. It is highly unlikely that you'll have a signal where in both bands the limiter would actually do something. Except if you use different thresholds in both bands.
If you want to limit a signal, do just that. If you want to do separate dynamics processing for different frequency bands, a limiter is probably not the most useful tool, musically. There are multiband compressors already, and they're of some use. Although even there it's already very hard to use them wisely.
You could just limit the two bands to [-0.5 .. 0.5] and you'd be sure you never overshoot the final signal. But still it would be unclear if such a thing is even desirable and why.
The thing is, if you split a usual music signal in two bands and limit separately, the biggest contribution to the peak level will be in the low band. It is highly unlikely that you'll have a signal where in both bands the limiter would actually do something. Except if you use different thresholds in both bands.
If you want to limit a signal, do just that. If you want to do separate dynamics processing for different frequency bands, a limiter is probably not the most useful tool, musically. There are multiband compressors already, and they're of some use. Although even there it's already very hard to use them wisely.
