Voxengo Premium Membership - All Voxengo Plugins For a Fixed One-Time Fee
Forums     Plugins     Redunoise Transition band semantics ??


Hi Aleksey,

When I set the 'Trans Band' param am I actually setting the width of the transition zone ?

The manual notes that the 'transition band is centered around the Threshold curve you have defined'.  Since there is only a single transition band control it must be controlling the width in dB of the transition band.

In other words (looking at the spectrum - from the top down in dB):

Above Transition Band (no reduction above this point)

Top of Transition Band (begin gradual attenation for signal falling below this point)

Middle of Transition Band (point where Threshold is set, medium attenuation)

Bottom of Transition Band (maximal attenuation)

Note: If Transition Band is set to 0dB this means its' width is 0dB and all of the items listed above occur exactly at the Threshold.

So if I have a Threshold (flat for simplicity - in actual use it will usually be the shape of the noise spectrum at the noise floor) at -40dB and a Trans Band at 10.4dB then the center of the Transition Band is at -40dB, the top is at -34.8dB and the bottom is at -45.2dB.

When the audio signal for a particular band falls to -34.8dB then a small amount of the 'Reduction' gain will be applied to that band.  Later if the signal drops to -45.2dB then the FULL amount of gain reduction will be applied to that band.

This assumes a History buffer of 1 , that's another story !  Basically an averaging for that band over time to 'smooth' the affect of dropping thru too many gain reduction levels (via downward expansion) too fast or too often and causing audible artifacts.  Does History include adjacent bands also or would that be a different setting ?

Do I have this right ?

Thanks,

kylen


Just to follow up on the History buffer setting and it's affect on adjacent eq bands.

I guess a side-effect of a longer History buffer would be that statistically over time there are bound to be a greater number of adjacent frequency bands being excited for the 'length' of the History 'window'.  Assuming the program material is not a single tone ! :)

Having a longer History window is just as likely to have an effect on non-adjacent eq bands also.

I don't know where I'm going with this except maybe I was worried about a single eq band having too much downward expansion applied creating a steep slope and 'swirlies' or 'flangies' types of sounds.  My experiments show that the History window eliminates this (among other things).

So far using Redunoise I'm getting results comparable (even better I guess) to a product I paid 5x more for.

kylen


Kyle, your first post here describes exactly how Trans Band works.

History parameter allows to get a smoother overall spectrum Redunoise uses during threshold analysis.  So, for small History values it is useful to set wide Trans Band values.  Also note that large History values can give some kind of EQ 'smearing' which results in some parts of thesignal being reduced for too long (this, in turn, eliminates swishes).

So, I guess the whole thing with History is a right balance between smearing and swishing.


Hello again Aleksey,

Would it be possible in Redunoise to have a frequency readout that follows the cursor, similar to that in CurveEQ?  If one were able to identify the specific frequencies of a noise in another program (like CurveEQ), might it not be useful to then go to Redunoise and immediately pinpoint that area or areas?

I'm still working on Redunoise, so maybe I've missed the point somewhere.  It's been known to happen...

So far, I think the results with Redunoise have been outstanding.  Thanks once again for your great work.

John


OK, I'll add that 'freq readout' thing to my to-do list.  I'm going to release beta 1.1 pretty soon.

Thanks for the details of the History Aleksey - I haven't heard the smearing yet and I've had the History window open pretty wide - almost all the way.  I'll review my settings.

I have the threshold set to the shape of the noisefloor (which the learn or fingerprint function will do automatically when you release it).  The gain of the threshols is set just above the noise floor by a couple of dB.  I've got a fast attack and slow release and the trans band is about 30 if I remember.

Gain reduction is set at about -20dB but I've got that line tilted to emphasize the lows.

Anyway, there are lots of controls to fine tune and I can see a use for different tuning across the frequency spectrum for all the controls.

John - I'm doing noise reduction on cassettes now so I 'draw' my threshold (by adding points) to exactly match the noise floor - any spot where there isn't music playing but there is tape hiss.  Then I select all points of the threshold and raise it a dB or so.

I'm sure my technique will improve as we get more info on the interaction of the params and example settings.  Maybe a manual update !  Looks very promising so far !

kylen

This topic was last updated 180 days ago, and thus it was archived.  Replying is disabled for this topic.