Hi Wave Arts,
I noticed that when I use 6-bands and default all the controls (so there is no dynamics processing, but enable is on) the output is about 3 or 4 dB higher than when using single band. I’m thinking the internal summing might be off a bit.

I put an instance of RMS buddy before MultiDynamics and another instance of RMS buddy after in Sonar4. Using 1 band there is a clear difference in output (the output is correct with 1 band) than when increasing the band count.

Can you guys reproduce that?
Thanks,
Kyle

I will look into this and see if I can reproduce it.

OK - thanks rpmartino.

I even went down to 2 bands - in bypass mode and something is happening I don’t understand - I’m in the red for some reason. I updated to the latest, now I’m using Multidynamics v4.12 still see the problem.

Hi kylen,

When you say you are “defaulting all the controls” what is each band set to?

For example, when I set all the bands to have a lo and hi gain of 0, with threshold = 0 (basically no dynamics processing) it seems the output of the processed and unprocessed signal are the same (at least using my ear, I haven’t tried RMS buddy yet).

Hi rpmartino - here’s some better specifics since you’re not seeing the problem on your side. I’ll keep it short so we can sync up.

I’ll write this in the form of steps to reproduce the symptoms - it’s easier for me to keep track of and probably better for anyone trying to see what I’m seeing.

1) Insert pink noise into a track in Sonar4. Loop the track. Sonar4 track meter shows -3.3dB
2) Insert an instance of MultiDynamics (DX version). Delete all bands except 1. Make the following settings:
Lo Gain = 0db
Threshold = 0db
Hi Gain = 0db
Ratio = 1
Attack = 1000ms
Release = 1ms
Knee = Hard
Lookahead = Off
Xover = 30db/oct
Output Gain = 0db
3) Sonar4 track meter shows -3.3db, MD meters show same.
4) Add another band to MultiDynamics (it will have the same settings as the first band). Sonar4 track meter shows -0.9db, MD meters show same.

The output level of MultiDynamics has changed simply by changing the number of bands from 1 to 2.

Well, it should be spot on. I went to great pains to design a proper crossover network for MD with perfect amplitude summing. However, there is a certain amount of allpass phase effect and this can alter the peak amplitudes by changing the waveshapes. But the RMS out should be exactly the same as RMS in. Hmmm, I’ll look into it.

Bill

Hi Kylen,

I just did a bunch of tests in Sonar4 and also SoundForge 7 and couldn’t repeat your example. The levels are correct. I tried it with mono and stereo pink noise files.

What sampling rate are you using?

Also, none of the bands are soloed, muted, or bypassed, right?

Bill

Hi Bill - I’m using 44.1KHz 16bit pink noise. No bands are muted or solo’d in the previous examples. Thanks for looking at this. I’ll try some more tests this weekend, if you guys can’t recreate it I must be leaving out some information…

I like the MultiDynamic plugin and it’s features and sound by the way - I’m just trying to get the levels accounted for in Sonar4. Maybe I’ll try it in Adobe Audition 1.5 and Soundforge6 and see if I get any different results there…

Thanks for the help!

billg -

...However, there is a certain amount of allpass phase effect and this can alter the peak amplitudes by changing the waveshapes. But the RMS out should be exactly the same as RMS in.

I think that is exactly what I’m seeing Bill. When I insert an instance of MD into a track the peaks amplitudes change. This must be because of the special Wave Arts crossover you guys worked up.

Usually when I insert a plugin into a track and all elements are ‘zero’d out there is no change in any levels or sound, it is transparent in other words. But even with all controls off the MD crossovers are still working aren’t they! The change in peak isn’t alway convenient - I’ll have to remember that and get used to it. My other MBs don’t do that as I remember so I wasn’t expecting level changes here in MD and thought there might be a bug of some type - sounds like it might be an expected side effect of the design then ? It would seem to make sense to try and keep the rms levels stable - given a choice of peak vs rms…

At any rate I can see the same thing in Adobe Audition 1.5 and even the MD Dynamic response display shows the level changes depending on how many bands are added.

Hi Kylen,

I don’t think the allpass phase effect can explain the discrepancies you report. Seems more likely that you are seeing something strange, and we’re just not able to reproduce the same problem. I’m going to go through it again, repeating your example.

bg

OK, I take it back, it is the phase effect that is causing the peak amplitudes to increase. I wasn’t noticing it before because I wasn’t letting things run long enough. I used a 20 sec pink noise file with peak amplitude of -12 dB. Running it through MD with 6 bands but no dynamics active you will get some peaks above 12 dB. With one set of bands I got -11.4 dB (+1.6) and another I got -9.8 dB (+2.2). This is like a rogue wave effect, by changing the phase relationships in the pink noise, every now and then the waves at different frequencies line up in phase and a peak pops up. The RMS levels are spot on. This peak effect is nothing to worry about and can’t be avoided using IIR filterbanks. An FIR linear phase multiband dynamics processor could be built that would not have this issue.

bg

OK - thanks Bill. I’m glad we’re seeing the same thing anyway. Thanks for the additional info about the IIR crossovers I gotta put that into the old noggin and spin it around a little. -Kyle