Can you help with these audio files?
 

I have two audio clips from a wedding that I hope can be improved a bit. The recorder on the officiant wasn't running so this audio is being picked up from a more remote mic so there is the characteristic room echo. I know this is a difficult audio problem but if anyone can help improve these files a bit I would really appreciate it.

I understand the need for backup recording sources and the always remarked upon "retake filter" but this is something that I was handed and what is done is done.

Here are the links to the files in my DropBox account.

http://dl.dropbox.com/u/8731209/Sound%201.wav

http://dl.dropbox.com/u/8731209/Sound%202.wav

Thank you for any help you can offer.

Hi Jim

Are these the only files? And will they have to be intercut with files that were recorded with a different mic/distance?

I don't think there's too much that can be done and even if the echo can be reduced a bit it will almost surely affect the voice quality to some extent so if these are the only clips it might be OK but if they'll be combined with other better recorded material I'm not so sure.

Just checking.My main system is off for repairs but I have older versions of the S/W on this system and I'll be happy to play around with it, but I think the best that we'll get is some reduction of echo with some degradation of the voice.

Hi Jim, There are two other places where the officiant speaks. One is a reading at the podium which is miced. The other is the vows and rings where he is picked up well by the groom's mic.

I know the best I can hope for is somewhat of an improvement but even that would help a lot. I would be most grateful for anything you can do.

This wedding was at the Carmel Mission in California. Since it's a stone structure, the natural character of the sound there has a distinct echo. So the sound won't be out of character if there is some remaining echo.

I away from my system with the good speakers right now (laptop is useless), but in situations like this there are a couple things you can try.

1. Compression can give you more presence on the audio track
2. adding reverb on the dry track will wash out the sound, but cause it to be more consistent throughout.
3. Rent a CEDAR DNS1000. There is a setting on it that can be used to essentially remove reverb. Doesn't always work and it is probably going to be an expensive rental, but desperate times call for desperate measures.

--Ben

Yeah, I know the Carmel mission well - I had a house in Los Gatos for 23 years before moving to Tucson. We used to drive to Carmel for lunch quite often. And we always took visitors to the mission.

I'll fool with it over the weekend, but I'm not expecting miracles.

Not sure if this is much or any better...
 

but take a listen and let me know

http://dl.dropbox.com/u/17745879/Sound%201%20result.mp3

Or maybe this one may be a tad better - hard to say

http://dl.dropbox.com/u/17745879/Sou...t%202A.mp3.sfk

Thanks Jim, I have been working with it as well and my results were similar to yours. I cut off the frequencies above and below the vocal frequencies and also applied the Transient Designer plug-in for my Universal Audio UAD2 card. Others more skilled at it might be able to do more. Working with audio filters is both an art and a science. I don't pretend to be either.

What I did was along the lines Benjamin suggested - first a noise reduction in Izotope which got rid of most of the low rumble and then a compressor in Sound Forge. It seemed to me that most of the reverb/echo was following the louder speech - usually at the start of a phrase so I played with the compression until there was a less emphatic sound at the beginning of each phrase which also brought down the levels of the echo. The I mixed a little noise that I tried to get close in level to the echo o try and cover it up a bit.

Unfortunately, I don't think there's a silver bullet that woks magic on echo because it lies right on top of the speech frequencies.

I agree. On top of that, speech is more or less a continuous tone. If it were a momentary sound, a lot more could be done to it.

Thanks for taking a look at it.

Jim, it sounds as if the officiant was using a PA system, and your recording picked up mostly amplifed voice. Is that correct?

If so, I'd try something that might strike you as odd, and would certainly involve a bit of work.

I would first try to understand the freq. response of the PA system and the room resonances. To do that (if I had a lot of time on my hands) I'd try to recreate the recording situation as closely as possible. I realize you can't bring the people back into the room for absorption. Aside from that, I'd set up the same mic and recorder that were used to capture the "bad" clip you're dealing with, in exactly the same location. Then, instead of having someone speak into the PA mic, I'd play a minute of white noise through the PA system, and record that with the same rig that recorded the "bad" clip. I would abruptly cut off the white noise, and also record the decay of sound in that room. Then I'd go back to the studio and look at a spectrum analysis of that recording, to find an EQ curve that would flatten everything out.

I would then apply that EQ curve to the "bad" clip to get a flatter recording to start with. Then I'd proceed to try to deal with all the reverberation.

Does this sound at all helpful or interesting? Or perhaps too time consuming?

Jim:

You can also try giving these files to a recording studio. These folks work with audio all day long. I have done some research on this in the past and from what I recall, they charge on average about 80.00 per hour, at least in my area.