Again, I’m sorry for the delayed response and again thank you for the help/reply. Re: silence cleaner crash: I tired with a second, different (long) file and again it crashes every time. Maybe the problem comes from files that are very long in duration? It doesn’t crash if I select only a couple of seconds. Also the files that it crashes with are ac3, so from a troubleshooting standpoint maybe that will help. I tried it with the sample wave I posted and it works good for the silent sections, but the portion with music still has the crackle/buzz.
Can’t you just isolate the frequencies of the buzzing, and remove them? You will of course degrade the quality of the audio to some extent, I mean the audio you want to keep. But it might make things clearer
What I regard as the “buzzing” distortion (that occurs when the signal occurs and not otherwise) is in the same frequency range as most of the higher frequency range of the signal.
So it sounds too dull when you remove the “buzzing” frequencies.
That is why Robert suggested averaging the samples. This helps to reduce noise in the upper frequencies while keeping a reasonable amount of the useful signal.
Gale
Robert will have to answer but Nyquist plug-ins may use a lot of memory while processing the audio. The longer the selection, the more memory they use. That memory cannot be released until the effect completes.
Silence cleaner is only intended for the audio before the signal starts. You still need to enter
(defun noise-red (s)
(snd-avg s 50 1 op-average))
(multichan-expand 'noise-red s)
in Nyquist Prompt to deal with the “buzzing”.
Gale
The original file type should have no influence on the silence cleaner effect. Or are you working with an unusual sample rate?
How Long is the file that you try to process?
I’ve just tried it on a 30 min mono-file and it did ok. Perhaps you should disable the preview Option for longer operations.
After all, it is a mere experimental plug-in.
As the Name implies - it cleans the almost silent parts, the rest won’t Profit by it.
(I’ve seen, I’ve written nearly the same last time - sorry for that.
Maybe I should re-read the Topic…
meanwhile… Gayle has posted.
The silence cleaner works on all Audio (if it is under the threshold - not only beginning and end.
The noise-reduction script has an entirely different Approach, it evens out the whole Show (as koz would say), i.e. makes the waves smoother.
Now we can try that the code works only on quieter parts, where the noise is not masked sufficiently. This requires a Little “tour de brain”.
A few minutes later…
I have a strong aprehension that this Sound isn’t restorable.
I hear only crackles, no buzz (in the General sense).
My masking proposal was an entirely bad idea - The crackles are getting more intensive in the louder parts.
Buzz is normally a constant Sound with a certain amount of harmonics at equally distanced frequencies.
This DVD is simply a production failure - as far as I can judge from the tiny sample.
The averager works well, but the higher frequencies will be lost unregainable.
The blast of damage at 5.7 seconds (the buzz) does respond to notch filtering, but it’s a moving target. The beginning of the damage seems to hover around 200, 400, and 600, whereas the end seems to favor 250, 500, and 750. After applying all these notches, I still had periodic damage around 220 Hz. I got these numbers by counting noise pulses on the timeline and doing the math. It worked, but it didn’t help. No matter what I did, and no matter what the blue wave response, the theatrical presentation still sucked.
I would call this damage “organized brown noise.” Sort of a brown-noise garden club where everybody separates into their own floral specialty – roughly equally spaced. Impossible to fix.
And yes, I agree that the feedback at the beginning would respond well to a notch at the howl frequency.
Koz
@koz
I believe your answer got into the wrong Topic.
I guess you mean rather the one with the speech (Brasilian?)that has a Feedback and at 5.7 s the queer Sound that resembles a Marimba (or so).
Sorry, Ihaven’t the link up the sleeve right now.
Besides, Nyquist has a variable notch filter which could start at 200 and end at 250 Hz.
But thats rather cumbersome to program, since the next occurance has certainly other Parameters.
Oops. There are two similar questions, both multiple chapters. Sorry. Koz
Again, thanks to everyone who responded and/or fiddled with the file. I appreciate it.
Sample rate is 48.0 khz. The original file I’m working with is 77 minutes. The second one I tried is 62 minutes. They both crash. As far as preview option disabled, I tried with “listen to removed sounds” set to “no” (I didn’t see disable preview) and it crashed this way too.
You can say that again.
I’m just about ready to agree with this statement.