I video taped a public speaker who was wearing a microport. But I was surprised when questions from the listeners were allowed at the end. And since I didn’t plan for it, I only have the sound from the speaker’s microport and from the shotgun microphone on the camera. I’m trying to clean up the sound and I’ve tried for hours, trying out the various tutorials on this. However, the results aren’t good. Lots of warbling and even worse sound, although the noise is somewhat removed. I would appreciate if someone would spend a couple of minutes with the sound clip and let me know if it really can be made usable (and how). I’ve uploaded a 41 second portion of both the microport and shotgun soundtracks in a 3,5 MB file here: http://www.2shared.com/file/-3_DNeHj/Sound_files.html
The sound starts with a couple of seconds of sound from the public speaker, and then from 3-6 seconds is the background noise that I think is the cleanest in the clips. Then from 18-41 seconds is the sound that I’m trying to clean up. I’m guessing it needs both cleaning and amplifying.
I think that the best you will be able to do is to “duck” between the channels - that is, when the main speaker is talking, fade out the shotgun mic and when the person in the audience speaks, fade out the microport. You will end up with something like this, which though not ideal is comprehensible.
So we don’t know what she was wearing and searching Microphones didn’t help
Oddly, the ability to rescue either sound clip is about the same. Nobody will confuse the results with a sound studio, but I think we can make the audience voice at least understandable.
Alright, there might be some terminology I need to get straight. She was wearing a wireless microphone - that’s what I meant
Thanks for helping out - can’t wait to hear more.
It’s a good rule of thumb when you’re telling us about problems is pretend we have to buy what you’re using. What we’re actually doing is building your system in our head so to better diagnose the problem.
So that still didn’t tell us anything.
Do you know what the camcorder was?
You just care about intelligibility, right? It doesn’t have to turn into Glen Glenn Sound Studio.
Oh, and rar is not a native Mac compression/archive format, so I had to find and install software.
Thanks for picthing in. I’m not quite sure what you need e.g.the camera info for, but it was a P2 recorder. I know why the sound gave me problems (I knew that even while I was shooting it), and now I’m just trying to save it the best I can. Let me know what other info you need, and I’ll make sure to provide it. Sorry about the rar-format. I’m on a pc and use it every day - guess I should have zipped it.
Yes, I only need the sound to be ok - it doesn’t have to be great.
Steve: Thanks for the input - much appreciated. If I falter in trying to denoise it, I think I will do as you suggest.
Kozikowski: That is most helpful, yes. Also very much appreciated. The result is pretty decent when you compare it to the original. I will give it another go and try to follow you pointers. It’s the whole MO that I need to master. Thanks again.
I ask these things because if you knew everything about the problem you wouldn’t be here and we know nothing about the problem or the setup and it’s not unusual for us to tighten up the license plate bolt.
That’s a punch line of a joke. The car won’t start. My wife insisted on tightening up the license plate bolt. So finally after hours of ignoring her and it still won’t start, I tightened up the license plate bolt. Not only did the car start, the the radio that hasn’t worked in three years started playing.
It also tells us whether you’re paying attention.
OK, here’s how I did it (slightly abbreviated).
Effect > Amplify on the overall clip. The whole thing’s way too low.
Select a portion of the audience performance with room noise only and Effect > Amplify.
Effect > Noise Reduction > Get Profile.
UNDO, UNDO.
Select only the audience portion and Effect > Amplify.
Effect > Noise Reduction
24
00
300
0.15
(settings top to bottom This is all in Audacity 1.3.13.)
Effect > Amplify > OK. Now you got what I got. We talked about automatically amplifying before the Get Profile step, but I don’t know if that’s included in Audacity 1.3.13 or not.
And as before, there’s not that much difference between the audience taken from the two systems. That was the camcorder mic.
Is any of this at all helpful? You should mess with settings in Noise Reduction to see if you can do better. I did those on the fly and with a fuzzy idea of what the controls do.
Reduction is the loudness difference between the hero performance and the junk. You can’t just crank it all the way up because the reduction will start to damage the show.
Sensitivity is a new one. I’m not sure what that does.
Smoothing is the integrity of the spoken words. Too low and everybody sounds like Donald Duck. Too high and the overall noise reduction goes away.
Attack and Decay affect the leading and trailing edges of each word. Too long and each word will have a tail on it. Too short and the consonants at the front of each word will be come damaged.
Audacity 1.2 had “straight” noise reduction. No fancy tools. It sucked for most jobs because you had no way to gracefully, gently separate spoken word and desired performance from the effect.
That still leaves you with integrating the crystal clear speaker woman with the corrected audience responses. You might be there for a while, but this should give you a push.
By the way, just that you made it to the second page of responses means we think you can rescue the show. The dialog would have stopped a long time ago if you had a basket case.
Trebor: Thanks - would never have started tinkering with the notch filter, if you hadn’t mentioned it to me.
Koz: All of this has indeed been very helpful. Your input is valuable - your comment and hints are what makes me dive right in and try things out again, but this time with more of an understanding of how to reach my goal. I’m already seeing better results - can’t thank you enough. That goes for all of you.
