Getting rid of crackle

[Craig Maier]

Hi Jan,

The EZ (or any of the Impulse Filters) are non-linear systems. Thus, benefits can be had with multiple passes there-through. I would recommend multiple passes with lighter settings compared to the ones that you mentioned.

We always tried to get as much with the impulse filters as possible via multiple passes (and sometimes "reverse file" passes too -). But, ultimately, we always had a few "ticks" that we had to deal with manually.

BTW - - - I do not do so much manual "tick" removal with my own personal restorations, but for pro/commercial releases, we go through them with a "fine tooth comb" (very carefully - - - ). The reason is that our work is replicated ten of thousands of times - - - so it should be perfect.

Does that help? If not, IM me.

Craig

[Jan Myren]

re:

Thats fine, but maybe you can give an example on your multiple running process with different settings.
Also, do you have higher settings on the median filter than just 3?
AND, do you think it was a clever idea to change the FFT from 2048 to 4096 in the CNF? Then I did not have that kind of "echo in the background"-ghost that I had on using those settings of att:125, rel.250 on FFT 2048. Earlier I have used att.time of 50, but that leaves a lot of dig.artifacts, that I also like to reduce, you know....

best regards
jan

[Craig Maier]

Hi Jan,

I think that part of this question is "when does the definition of Crackle end, and the beginning of Hiss {or random Noise} begin.

There is a humanly percieved difference between the two, but mathematically, both involve random occurances. In the case of what most people call ticks or clicks, one uses the Impulse filters. If one is dealing with very tiny clicks or ticks, one uses the narrow crackle filter. If one is involved with extremely high density clicks (hiss), then one uses the CNF.

I think that if you follow the threads of "Doug" here on this site you will see a good methodolgy for reducing crackle to its absolute minimum without killing the top end of the audio spectrum.

I think that it is important to understand the difference between "In - Band Noise" and "Out - of - Band Noise". The reason is that you need to eliminate the "Out - of - Band Noise" before applying the CNF. How do you know what the range of frequencies involved in the "Out of Band" range are? Well, the spectrogram is a good place to look to make that determination. I think that you may find pics and desctriptions by Doug about identifying the "Out of Band" noise and how to "Brick Wall" them out of the equation - - - -


References should be able to be found here on the forum. If you can not find them, let me know.

Craig

ps - - - btw - note that the use of the CNF is part of the process of eliminating what many refer to as "crackle".