I have a larger big trouble concerning my wedding video. We got married on a beach and on that day specifically the wind and the waves were rather bad. I can hear about 98% of the word in the audio but it is heavily contaminated by wind and wave sounds. I have been trying to clear it up myself using some instructions but have not succeeded very well.
I understand that this is a big trouble as the wind/waves varies in db and in frequency, but I am not able to reshoot it so I am stuck with this. Do you have any tips for helping me or better yet if someone is able to take a look at the file themselves (I’ve added the wave file in here). If it gets 1% better than the current state I would be glad.
High Pass filter, 300 Hz, 48 dB per octave may help a bit.
The “Isolate Vocals” setting in the “Vocal Reduction and Isolation” effect helps to make the main voice clearer, but at the expense of creating weird bubbly sounds in the background (“strength” should be set to about 1.0).
Thanks! The high pass filter defidently helped sort out the highest areas. The isolate vocals defidently helped make the voices clearar but at the cost of making the sound surreal (The video of a couple on the beach and that background noice makes it a bit too weird to put in a wedding film)
I’ve tried to play around with other settings, is there some other action apart from high pass filter that you think can be applied to tone the wind down? My belief is that the volume of the voices are fairly similar, not possible to distinguish the noiser wind based on that? (or something). Thanks again!
Thank you very much for your input. The high pass filter was a good suggestion and although the isolate vocals did make the sound more clear, the bubbly sounds would not suit in very well with the wedding video with the ocean in the scenery. Do you have any other suggestions on how to further improve from the high pass filter? Did make it alot better.
If you apply a high pass filter, you should take the option “Isolate Center” and not vocals.
Apply it on a duplicated track and mix the two tracks together (by turning down the original) such that bubbly noises are still masked.
After all, you wanted only 1 % improvement…
perhaps, a bit noise reduction with one of the available noise types as profile may help too.
@Robert J. H., Do you mean copying the track so I have two then apply High pass filter on one and then isolate center on the other and then mix them? Do you have any recommendations on the settings for isolate center?
Well what I wanted was to have a crisp perfect recap of the voices and a slight lovely breeze in the background, but I would be glad even for 1% improvement which isn’t to say that I will settle there
@DVDdoug, Thank you I believe that you are correct. I will try my best to improve the quality of the audio but then add subtitles. Also makes it more genuine compared to just adding a voice over.
“Remove Vocals” has a built-in high pass filter (look for “Low Cut” so you don’t want to use a second one.
However, I would still isolate the center and apply a high pass afterwards because the built-in filter is very sharp and causes some resonances for this special audio track.
The main problem we are facing is that the three voices are not exactly in the middle and the left and right channel are overall not much related. In fact, they have a correlation of -6 percent!
One way to use isolation is to remove only the very extrema in both, hard left and right. A strength of zero does this.
The bubbling sound will be much less.
You could mix in some other sea/wind sound to mask it further.
Cheating? Yes, it is, but any effect (except amplification) is cheating.
You could also overlay ocean and beach sounds almost like this was intentional—in addition to the subtitles.
We still can’t take a performance apart into individual performers, voices, instruments and other sounds. Once you jam everything together into one sound file, it’s pretty much the end of the world.
The Big Kids occasionally get stuck with a bad shoot, and since there are mega-bucks on the line, they do ADR/looping. That’s where the performers come in and watch a segment of the video or film again and again in a quiet studio and speak new, clean lines repeatedly until they line up with the original screen action. Then on to the next segment. The original damaged shoot generally isn’t used except maybe for background effects.
You can schedule ADR with people who sound similar to the original people.
You could also try an on line noise reduction tool such as: http://audiodenoise.com/
Or let the audio run through Skype…
No tool is perfect and you will have to puzzle the best pieces together.
Also, some compression will be needed as well, the wind blasts reduce the volume, especially in the highs.
If this is from a camcorder, then yes, there is Auto Gain Control running trying to make all the sound arriving come out to an enjoyable show. The instant the wind increases enough to be dominant, the AGC “throws away” the actual voices and tried to make the wind into the show. So even if we could rip the show apart into voices and sounds, the voices would take very serious hand-processing to make them even close to being enjoyable.
Without the AGC, you have a different problem. The wind pushes the voices into overload where without the wind, the voices would have been fine. Overload is non-recoverable.
Thank you very much for your input.
Cleaning up audio is quite difficult, especially as it requires you to ask yourself “Better or worse” with 50 different combinations (keeps me thinking of Family Guy’s "Yafar may need glasses: https://www.youtube.com/watch?v=FLXmyau8Tmc)
Based on your input I’m quite certain that you could do more, but I’ve tried playing around with the settings and am not able to get a good sound delving deeper in to isolate center etc. So I think I will settle with high pass filter (300mhz,48db) and then to keep the original soundtrack with -20db just to keep some of the waves, add subtitles and then do voice overs on the absolute crazy bad part when me and my wife does our speeches to eachother (about 2 min in Part I which I didnt send out before).
That should make it quite watchable although with a fair amount of noice
If you have the time/inclination please have a look and review if you think any direct actions that I could take, otherwise I would like to thank you for your time
As I said, I would mix the center-isolated version (Strenghth 0) with the original, this reduces the extreme clicks on the sides by at least 6 dB.
Also, I wouldn’t go as high as 300 Hz, that sounds artificially to me as well.