• Wondering which camera, gear, computer, or software to buy? Ask in our Gear Guide.

Question about audio clipping.

H

harmonica44

I have dialogue that was recorded dry, that I want to add some filter and reverb too. The dialogue does not clip and does not reach higher than -6 db. But after I add some filter and reverb to it, it clips, even though the waves picture, clearly shows that it is not clipping and has not been recorded that way. How can I modify the track, without causing it clip from the effects added?

I tried turning down the amplification, and even though the picture does not show any clipping, it still sure as hell sounds like it's clipping.
 
Last edited:
Sort by date Sort by votes
Convolution reverb and FFT filter. Essentially I am trying to do things like make dry dialogue sound like it's coming through speakers in a room, than with the reverb of the room added to make it sound like the speakers are being played in that room. I am using Adobe Audition.

It actually sounds convincing accept for the fact that it clips really badly.
 
Last edited:
Upvote 0 Downvote
The input or output levels of the "filters" - they are not filters, BTW, they are PROCESSORS - are probably too hot. As you add signal to (process) the sound you also increase the level - you're ADDING to the signal right? So, your original sound clip is at -6dB; you add some low end and some upper mids and now the output level of the clip is OVER 0dB, hence the distortion. Or, you are just at zero coming out of processor #1, but as you add signal in processor #2 you push the level over the limit.

See? Gain-staging applies to audio post as well as production.


If you had actually read all of the books you say that you have, and actually read the responses to the posts by myself and the other audio people here on IndieTalk you would already know this...
 
Last edited:
Upvote 0 Downvote
Well I don't recall reading that or if I did, I just did not understand it till now. In Percel's Dialogue Editing, the book does not go into that in any part I can find.

So what is the solution then? If I turn the audio down before applying the filters, then it's just too quiet and too much of the noise floor is brought up too high.
 
Upvote 0 Downvote
I don't know anything about filters; I never use them.

Most PROCESSORS have input and output levels.

pt1pteqiii.l.jpg


cw_m700xm700_altiverb.jpg
 
Upvote 0 Downvote
Funny AA, those all install under the "Audio Filters" menu in Final Cut ;) Nomenclature aside, anytime you alter a signal, it has the chance of changing (Mathematically speaking that rhetorical method would be represented as A = A). Any change can have unexpected results if you haven't played with the Processor (filter) enough. Keep an eye on the meters, waveforms and the spectrographs after you apply each filter (sorry: processor -- which I guarantee is the correct term). Since you have limited experience with audio post, the meters and tools will guide you as you learn to visualize what they say in your mind's eye.
 
Upvote 0 Downvote
Okay thanks. I am finding that as I go. Another thing is let's say I have a guy say two sentences. Both sentences are no higher than -6db. I then brought it down to about -24db, then added the filter and reverb.

After doing that one, of the sentences clips and one is back to to -6db. Why is it that one is a lot higher than the other when they both recorded at -6db to begin with?
 
Upvote 0 Downvote
How similar are the lines? Is one recorded closer than the other to the mic? Different aspects of the sound are accentuated when you use a convolution reverb in the same way one room will reflect lower frequencies more than another. Maybe one of those clips has more bass?
Another possibility is... Are you using "peak" or "RMS" to judge levels? "Peak" means from one sample to the next, what is the volume? RMS is an average over time and used to more precisely judge how we interpret loudness naturally. Software on the other hand, will distort the sound if even a peak is too loud.
 
Last edited:
Upvote 0 Downvote
Are you fookin' paying attention?

Alcove Audio said:
The input or output levels of the PROCESSORS are probably too hot. As you add signal to (process) the sound you also increase the level - you're ADDING to the signal right? So, your original sound clip is at -6dB; you add some low end and some upper mids and now the output level of the clip is OVER -0dB, hence the distortion. Or, you are just at zero coming out or processor #1, but as you add signal in processor #2 you push it level over the limit.
Click to expand...
 
Upvote 0 Downvote
Yeah I get that, but I you did not mention how to fix the problem. What do I do instead? I know thier is no magic button, but their must be something that can point me in the right direction.
 
Last edited:
Upvote 0 Downvote
harmonica44 said:
Another thing is let's say I have a guy say two sentences. Both sentences are no higher than -6db. I then brought it down to about -24db, then added the filter and reverb. After doing that one, of the sentences clips and one is back to to -6db. Why is it that one is a lot higher than the other when they both recorded at -6db to begin with?
Click to expand...

Simple question, where are the two sentences peaking at -6dB to start with? A peak reading represents the highest value in the frequency spectrum of the signal you are measuring. Let's say sentence "A" has it's peak (-6dB) at 500Hz and the other, sentence "B", has it's peak (-6dB) at say 800Hz. Now apply EQ to both sentences, say a 6dB boost at 800Hz. Sentence "B" is going to hit 0dB and clip, while sentence "A" isn't! Adding reverb or pretty much any processing effect is going to affect some frequencies more than others and how much it will change the overall peak level depends on the frequency content of the recorded material you are processing. If the highest value of 800Hz in sentence "A" is -13dB and we add 6dB at 800Hz the peak level reading is not going to change because the peak value at 500Hz is still greater: -6dB at 500Hz as opposed to -7dB at 800Hz.

John Purcell's book is an excellent book on Dialogue Editing, the best you can buy but as the title suggests it is about Dialogue Editing. It is not so much about dialogue mixing/processing and contains very little audio theory. You need to read some books or (online forums/articles) on basic audio theory!

G
 
Upvote 0 Downvote
Okay thanks, that helps! I've tried this before another time, but it seemed to bring up the noise floor. Is their a way around that? Or is their a limit to how many dB you can add, or take away, before the quality goes down?
 
Upvote 0 Downvote
Mmmm, I don't think it has helped. If it had, you wouldn't have asked that last question!

The reason your two sentences behaved differently was because they have different frequency content, as I said before, it "depends on the frequency content of the recorded material you are processing". Therefore: The limit of dB you can add before the quality drops depends on the frequency content of what you are processing and of course "quality" is a relative value judgement anyway. How much the noise floor is raised depends on what frequencies you are boosting and the frequency content of the noise floor.

G
 
Upvote 0 Downvote
Okay thanks. I tried changing the levels in Audition and brought them down before adding the effects. But now I can't get it loud enough without it clipping again afterwords. After the effects are done you have to turn it back up and that still causes it to clip.
 
Upvote 0 Downvote
You must log in or register to reply here.
Back
Top