|
Author
|
Topic: Effective Decrackling
|
younglove
Member
|
posted 06-11-2000 02:26 PM
   
Here's a technique for effective decrackling
of recorded vinyl. It's good for vinyl
recordings that have constant low-level crackling and small clicks throughout. It
removes the crackle and clicks with much less
harm to the audio than regular declicking.1. Collect (paste) together all the parts
of the vinyl that have no audio, such as the spaces between tracks. 2. Get a noise sample over this material. 3. Perform a Noise Reduction over the original recording using this noise sample
using 100% (maximum) reduction, Keeping Noise Only (I use 400 snapshots, Precision Factor 19). *Save this into a clipboard.* This will collect most crackling and small clicks, along with vinyl surface noise and a little signal audio. The idea is to get a file with most of the crackle/small clicks and little of the signal audio from which to perform the actual declicking. This way, the declicking will have little signal audio to work with that it could remove along with the crackle/clicks, so that there will be much less signal audio degradation than with regular declicking of signal audio. 4. Declick the noise file using a healthy declicking setting, to detect and fix most of the crackle/clicks. Don't worry as much as usual about harming the audio, since you are not declicking the original audio, just the noise/crackle from it. 5. Paste-Overlap-Invert the noise file *that you saved in a clipboard in step 3 above* onto the declicked noise file, yielding just the (inverted) crackle/small clicks captured in the noise file. 6. Paste-Overlap the inverted crackle file onto the original signal audio file, yielding a decrackled signal audio with almost no degradation of the original audio signal.
For larger clicks and pops, you can now use a more conservative declicking setting that will catch more obvious clicks (that the decrackling method above will not get), but will leave the signal audio alone, or declick manually, for final touchup. [This message has been edited by younglove (edited 08-18-2000).] [This message has been edited by younglove (edited 09-29-2000).] |
DonR
Member
|
posted 07-31-2000 10:13 AM
Hey, this really works! I've used it on about six LP sides now, and it's become my routine first step. It effectively removes the crackle and small clicks with no noticeable degradation of the audio. On some recordings with severe crackle, I run through the procedure twice before any other processing.The really nice thing about this procedure is that, while it takes a lot of time (careful vinyl restoration always does), it's a "cook-book" procedure that doesn't require a lot of trial and error, thus doesn't need much of MY time, just the computer's. I find that even on a very noisy recording, what's left after one or two passes of this procedure is easy to clean up. Perhaps a mild noise reduction pass, then manual de-clicking for the remaining clicks and pops. Thanks for posting this! |
Joe
Member
|
posted 07-31-2000 10:20 AM
Make a lot of sense! I'm in the mood now to dig my old Kraftwerk and Wendy Carlos records out of the basement...Curiously Refreshing Music: http://mp3.com/alectroboy |
Thomas B
New Member
|
posted 07-31-2000 11:39 AM
Thanks younglove! Why havenīt I thought off this before :-) |
beetlefan
Member
|
posted 07-31-2000 01:24 PM
It still sounds complicated but i'm going to try it on an old Jackson 5 and an old "Frogman" "Henry 45. |
resistor man
Member
|
posted 07-31-2000 10:43 PM
Most ingenious! I salute you. |
younglove
Member
|
posted 08-18-2000 09:25 AM
Title: CEP 1.2a Decrackle AS-3 44100
Description: Paste together all non-signal parts of wave,
Description: get a noise profile from this non-signal wave
Description: and make sure to save it into C:\coolpro\scripts\decrackle.fft .
Description: For this script, the noise profile must be taken from
Description: a 44100 wave file, and the wave file to be decrackled
Description: must be 44100. The selection to be decrackled must be large
Description: enough to accomodate the FFT size used in the NR command.
Description: Run script on original wave.
Description: May turn OFF Undo
Mode: 4cmd: Channel Both cmd: Copy
1: 0 cmd: Noise Reduction\Noise Reduction
1: 1
2: 0
3: C:\coolpro\scripts\decrackle.fft
4: 8192
5: 400
6: 19
7: 44100
8: 2
9: 32
10: 8
11: 9.42576e-011
12: 0.00222947
13: 1.75127e-010
14: 1.78127e-005
15: 1000
16: 0
17: 0
18: 2
19: 2
20: 0
21: 1000
22: 16384
23: 1000
24: 0
25: 1
26: 1
27: 0
28: 0
29: 0 cmd: Copy
1: 1 cmd: Noise Reduction\Click/Pop Eliminator
1: 40
2: 32
3: 10
4: 16448
5: 12
6: 0
7: 65
8: 0
9: 0
10: 0
11: 6
12: -11.4
13: -85
14: 3
15: 1
16: 6
17: 0
18: 30
19: 55
20: -40.1
21: 0
22: 0
23: 0
24: 0
25: 0
26: 0
27: 0
28: 0
29: 0
30: 0
31: 0
32: 0
33: 0
34: 0
35: 0
36: 0
37: 0
38: 0
39: 0
40: 0
41: 0
42: 0
43: 0
44: 0
45: 0
46: 0
47: 0
48: 0
49: 0
50: 0
51: 0
52: 0
53: 0
54: 50
55: 0
56: 0
57: 14
58: 20
59: 0 cmd: Paste Special
1: -1
2: -1
3: 3
4: 1
5: 0
6: 0
7:
8: 0
9: 1 cmd: Paste Special
1: 1
2: 1
3: 3
4: 1
5: 0
6: 0
7:
8: 0
9: 0 End: [This message has been edited by younglove (edited 09-29-2000).] [This message has been edited by younglove (edited 12-01-2000).] |
luezuve
Member
|
posted 08-18-2000 09:35 PM
Two words. Or maybe it's one.Woo-Hoo! LueZuve |
beetlefan
Member
|
posted 08-18-2000 10:25 PM
I'm not sure of how to save Ypunglove's script to a script file.When I thought I had it loaded I ran the script to test it. All it did was copy my sample wav file and stop. What's going on here? |
younglove
Member
|
posted 08-18-2000 11:31 PM
Beetlefan, the script posted was my 96000 Hz
version. It won't work with 44100 files. I
edited the posted script to work with 44100 files. For files of other sampling rates, edit the script to change all (5) occurances of 44100 (96000 old) to the new sampling rate. |
beetlefan
Member
|
posted 08-19-2000 03:14 AM
Younglove, even with the revised script for 41000 I can't get it to do more than copy. After it copies the file it just stops with an on-screen notice that the script has stopped. I couldn't quite understand your instructions so I gave up and started working on my own methods. At least I figured out how to get rid of clicks and pops without damaging the music. I got the idea from you. I just copy the original file, declick with ClickFix with saving clicks only checked, inverting that, and either pasting over the original file or mixing it down with the original in the multitrack. It is simple, invert the polarity of one half and blend it in with the opposite. One cancels the other out! duh. I tried to do it with the noise but it didn't work too well. [This message has been edited by beetlefan (edited 08-20-2000).] |
ChristianB
Member
|
posted 09-13-2000 07:29 PM
yOUNGLOVE Hi. Your posting for effective Decrackling is outstanding.
How in the world did you make this discovery?
What is your proffeshion?
Just who are you?
How old are you?
Why did you not patent this?
Software giants such as Sound Forge, Stein Berg, Along with any other software giant Could certaily use a lesson from you.
Thanks this is just remarkable.  |
jsm
Member
|
posted 09-16-2000 05:39 PM
younglove
If this turns out to be as good as theyr all sayin i'll luv you forever m8
I'll feedback when I get to try it out...
till next time... |
younglove
Member
|
posted 09-27-2000 06:37 AM
ChristianB, I'm a software engineer by profession. I do audio for fun. Does your question mean that you have some audio work for me? Anybody else?? |
Parky
New Member
|
posted 09-27-2000 11:49 AM
Hey Younglove how are you at making distant voices sound as if they are close? Or closer maybe? |
younglove
Member
|
posted 09-28-2000 09:20 AM
Parky, please email me. Thanks.[This message has been edited by younglove (edited 09-28-2000).] |
mr gerbik
Member
|
posted 09-28-2000 01:52 PM
Can anyone tell me how to load a script? I want to give younglove's method a go - what do I do with the script in your post? |
vishv
Member
|
posted 09-28-2000 08:02 PM
I couldn't get Youngloves script to work,
but I posted a script a few days ago to this fu\orum which should work |
younglove
Member
|
posted 09-29-2000 09:03 AM
I modified the script above. Read the description in the script. |
ChristianB
Member
|
posted 10-05-2000 07:41 PM
Younglove, Hi. Could you please tell the settings you are using in CEP1.2a as a goodly amound for Declicking the kept noise sample?
for example: settings for sensitivity
Discrimination
Detect big pop
Multiple passes
Pop Oversamples
Run Size
Thanks.  |
younglove
Member
|
posted 10-05-2000 09:57 PM
Well, the point of the decrackling method
is that declicking from noise effective for decrackling and is less harmful to signal audio than regular declicking alone. Settings for declicking and noise reduction
can be modified as per usual for different results with different sources.I haven't really attempted to optimize very much in the settings in the script above for decrackling, but I just wanted to offer a script which you can try and adjust yourself. So, my settings, which are only soft recommendations, are to turn on "Smooth Light Crackle", turn off everything else except perhaps "Multiple Passes". I used an "Auto" FFT size mainly because I haven't really experimented much with a general optimal setting, pop oversamples of 10, and run size of 12, Detect max, avg, and min values of 40, 30, and 6, and thresholds of
-11, -40, and -85. But you may well come up with better values your sources. |
DanT
Member
|
posted 11-30-2000 12:23 PM
First, I want to say that this method works quite well and is a real time saver. Thanks for sharing it with everyone.I am having a little trouble with the script however in CE2000 v.1.1. I have two questions. One, the script seems to actually paste-overlap the original wav file onto the inverted crackle file for step 6 (last Paste Special command in the script). I'm assuming that the result is the same as it would be if you were to paste-overlap the inverted crackle file onto the original wav, as the text instructions indicate. Is that correct? Second, if I allow the script to run without the prompts it doesn't work right, while it will work properly when I run with prompts. Anyone else seen that? Separately, but related, in CE2000 v.1.1 the script processor seems to ignore the clipboard argument to the paste command (straight paste) and pastes from the "current" clipboard -last one set by a previous command. Anyone with an idea on why? Finally, does anyone know of a CE script language reference? While capturing commands entered interactively into a script file is a really nice feature, it would also be nice to be able to write scripts. It would be even nicer to know, for example, what argument number 24 to the Noise Reduction\Noise Reduction command meant. |
younglove
Member
|
posted 11-30-2000 09:45 PM
> I'm assuming that the result is the same
> as it would be if you were to paste-
> overlap the inverted crackle file onto the
> original wav, as the text instructions
> indicate. Is that correct? Yes. > Second, if I allow the script to run
> without the prompts it doesn't work right,
> while it will work properly when I run
> with prompts. Anyone else seen that? I haven't. But I don't have CE2000 1.1. > Separately, but related, in CE2000 v.1.1
> the script processor seems to ignore the
> clipboard argument to the paste command
> (straight paste) and pastes from
> the "current" clipboard -last one set by a
> previous command. Anyone with an idea on
> why? You found a bug. Always use Mix Paste (Paste Special) in scripts instead of Paste.
You should never need straight Paste. > Finally, does anyone know of a CE script
> language reference? None exists that I know of. You just have to experiment. To see what the arguments are, record two scripts of the same
command, changing only one thing, and notice the difference in the resulting script. Some commands have repeating groups of arguments (like for each point in a graph), so that argument 54 may not always mean the same thing. |
gunars
Member
|
posted 11-30-2000 10:22 PM
I'd like to second DanT's request for script information, preferably from the good folks at Syntrillium. Besides the arduous task of reconstructing the parameters by experimentation, some of the values are still a mystery to me. For example, why do my recorded scripts have a line like: Selected: none at 1063619 scaled 2181207 SR 32075 when I don't have anything selected. I expect that's what the 'none' is for, but what are the other three numbers? Also, at least some of the procedures appear to have a different numbers of parameters between CEP and CE2k. A clear description of the values would help us modify CEP scripts to run on CE2k and v.v. Anyone at Synt listening? |
gunars
Member
|
posted 12-08-2000 07:31 AM
...or anyone else for that matter...  |
gunars
Member
|
posted 12-27-2000 09:14 AM
WARNING: real question resurfacing above the recent muck... |
UBBFriend: Email This Page to Someone!
Forums
