Sonnox learning time.

I understand that the Sonnox Restoration plug-ins may need some time to “learn” the material they are processing before they can be fully operable.

I have been restoring an old vinyl LP and have checked and adjusted the three plug-ins (De-Noiser, De-Clicker and De-Buzzer) over a specimum of about 20 seconds of a track. My view is that this 20 second section is a good representation of the whole LP, so the settings should be similar for the rest of the file.

I noticed that when I have the De-Clicker set up on a looped 5 second file properly with the plug-ins working and thus no clicks audible, that when I come back to it with the same settings it took 4 or 5 passes before the clicks became inaudible again. I assume that means it needs 5 passes or maybe simply it needs 10 - 20 seconds to learn the file.

In any event, I tried to render a file (a full 3 minute track of the LP) with all three Sonnox plugins active, and Wavelab came up with a message saying that “the De-Buzzer will take time to learn the file and the beginning may not be dealt with fully. Please use Freeze”. The message did not mention De-Clicker or De-Noiser.

So, how do users approach this? Obviously I have to use freeze on the De_Buzzer, but should I simply put all three plug-ins on Freeze?

The reason why I think this may be an option is that, as I said earlier, I am working on a recorded vinyl LP and I believe that the setting on that part I checked will be the same for the whole LP, as no parts are particularly worse or better than the part I have assessed.

So, the bottom line of my question is : is it best to use Freeze if I think I have already selected settings for a small section that will do for the whole file?

Please excuse me if I have misunderstood anything because I am new to restoration and the Sonnox plug-ins.

You should read about this batch processor plugin:

Thanks for the response, PG. I did not understand the graphic in your link at first, as I have not heard of a Monopass or Instructor. However, I searched around and read page 819 of the Wavelab manual, so now I know what a Monopass Instructor is. And, yes, I think that could help.

However, I think the Batch Processing Tools are only in the Full version of Wavelab 8, and I have Wavelab Essential 8. Am I correct in that assumption, and if so, without the full version, what can I do regarding my question about using Freeze?

I wonder if anyone else can add anything here? Certainly I don’t have the software PG mentions, but if my assumptions are correct about the Sonnex Restoraton tools needing to “learn” the audio first, then, from reading up on audio restoration, I think there may be a couple of solutions.

  1. Get the right settings, (using a number of passes), then simply set it on Freeze. (I think that Freeze takes those settings and retains them throughout the restoration process - but I could be wrong!)


  1. Copy some of the audio file and add that copied part onto the front of the file. In that way the tools will be able to spend 20 seconds or so “learning” the audio. Then leave it on Auto/Adapt during the process. Then after render I can simply cut off the previously added first part to get my full restored file.

I also read somewhere (though I cannot see it in the Essentials manual) that a De-Clicker needs to be used first. Is that correct?

Can anyone comment if have the right views? I would like to finish off the project so I would very much appreciate any help.

Hmm, no responses. Guess I’ll have to figure it out for myself then!