Improving Audio Quality - Voices

richardfrost wrote on 3/30/2005, 6:18 AM
My son, who is 12, is the film maker in our family. I edit our family home movies, but he makes features. To date, he has just done stop motion, based on lego, to which he then adds voices. You can see his work at http://rstudios.brickfilms.com

He has recently switched to live action movie making, and has fully scripted his new movie. We know that, if in future live action is what he will be doing most of, we need to be use lapel or boom microphones.

Despite that, because this is his first movie, we spent the day yesterday in a local forest doing all the filming, knowing that recording the voices using the on camera microphone would be a problem. Especially so as there was a river nearby and some of the filming was near a waterfall.

Question One:
What would you guys recommend as the best way to enhance the audio so that the voices come out more clearly against the general forest/river white noise type background.

I'm looking for software tips and/or best settings. We currently use Audacity for sound work but are happy to consider alternatives, so long as the cost is within the budget of a 12 year old's pocket money!

I can post audio / video samples if required.

Question Two:
Should he decide to do more live action work, what would be an adequate budget set up for microphones.

< Off topic anecdote> Whilst the 12 year old was filming, my 9 year old was acting. Whilst being 'unconcious' on the ground, he noticed an ant walking across his chest. After I calmed him down and reassured him there was just the one ant, he finished the scene. Just after that, about ten feet from where his head was, I noticed the biggest ant's nest I have ever seen. There was about a trash can lid sized black mass of myriads of Northern Hairy Wood Ants. I got some great close up footage if anyone is interested!

Comments

Chienworks wrote on 3/30/2005, 7:54 AM
Richard,

You won't find much that can help. Maybe some EQ will let you narrow it down to the key vocal frequencies, but that's about it. The noise of the river is full frequency range and varying, so Noise Reduction won't help.

May i suggest you take this opportunity to dabble in ADR? This would involve having the actors re-record their lines back at home in the studio. Vegas is very good at this task. Even the younger kids can get the hang of speaking their lines again in sync with the video on the screen.

Feel free to post videos at vegasusers.com
richardfrost wrote on 3/30/2005, 10:55 AM
Thanks for that.

I posted that whilst I was at work. Meanwhile, my son was at home dabbling with revoicing the tracks. What we have decided to do is copy the sound track to another track, then edit out the original voices from the original track and replace the gap with another section of the forest noise, then drop the level of that track.

Each character's voice will then be recorded on separate tracks, using the copy of the original voice as a guide. The copy track will then be deleted.

Is this the best way?
Jimmy_W wrote on 3/30/2005, 11:19 AM
Thats a tough way and it most likely look like a kung fu flick.
IMO

Jimmy
B.Verlik wrote on 3/30/2005, 12:16 PM
and please use Vegas for your audio. Audacity is okay before Vegas, but Vegas is much easier in the long run and has much more to offer. (just right click and add another audio track. As many as you want) You can do all the video and audio in Vegas.
farss wrote on 3/30/2005, 1:33 PM
Vegas can do ADR very nicely and it'd be good experience for your son as this is how 80% of dialogue is done in movies.
Put a clip on the T/L, mark out a loopable region around it.
Add an extra track, arm it for record, get an external mic into the PC (obviously), feed the mixdown to headphone.
When you hit the master record Vegas will playout the original audio while it records the new and the talent gets to hear both and you can control the mix to make it easy for them. As Vegas records each loop is a separate Take.
Very, very easy to do well in Vegas.
Sorry if my explaination isn't exactly step by step but hopefully you'll get teh idea well enough to realise it's a cake walk.
The MUCH harder part is making the new voice sound like it belongs in the scene. Don't you love it in those C grade action movies where the talent have been fighting for five minutes. leap up and say their lines without a single bit of breathlessness.
Bob.
Rednroll wrote on 3/30/2005, 2:15 PM
"Is this the best way?"

Yes, this is the best way. In my opinion your method is perfect. It will give you the most naturaly sound as it was recorded in the original environment, yet give you the quality for the voices you also want to achieve. The only tough part will be to get any camera seen, which has a face shot that must sync up with the picture. This would be a good use of using the Vegas Loop record function, where you can have the scene and the prior record loop at the same point while your son keeps saying the same line. Then you can toggle between the takes and see which one fits the picture best and then even composite edit them if you have to.

I've done many TV commercial spots, and in the advertising business the ad agency is always wanting to update something the actor said. Well, in some instances the original recordings where shot outside and now here we are working in a quiet studio trying to change a "Word" or a "phrase" and have it match up with the original recording. The method you outlined is exactly what we do. If you happen to see any Scott's Lawn TV commercials this spring. Those spots have been airing for over 4 years now and the original takes where done outside on the actors lawns. There's 3 different ones I've worked on and all of them have had line replacements......listen for them and tell me if you can hear them if you happen to see them. :-)
Jimmy_W wrote on 3/30/2005, 2:34 PM
Good info Red I think I will try that myself. I could use this on parody type stuff.
Jimmy
logiquem wrote on 3/30/2005, 6:04 PM
>You won't find much that can help

I use mysefl all the time SoundForge + SF NR to precisely filter unwanted background from nature sound samples (from bird for example). It works very well, altough it is certainely not in your budget.
Coursedesign wrote on 3/30/2005, 6:20 PM
Logiquem,

Were you ever able to filter out crickets behind dialog?

I was shooting an interview with an older lady from Peru in a rural valley, when the crickets got going with phenomenal energy. We could not go anywhere else, and just had to do what we could with what we had.

In post I worked pretty hard with NR and had a hard time. Finally got something that was useable, but it wasn't my proudest moment.

In retrospect, I wish I had had a discreet tan near-invisible Countryman B6 with an over-the-ear clip for that shot. With that, the mic element would have been millimeters from her mouth and the crickets far away... Even better would have been a noise-cancelling mic that would have automatically nulled out the common mode noise, but those are bigger.

Rednroll wrote on 3/30/2005, 7:01 PM
Regarding your cricket delima, of course not having it there in the first place is the best scenario. Another useful tool for something like that is an audio spectrum analyzer. I'm not that familiar with what the Vegas Video scope is for, so correct me if I'm wrong. I think what the video scope does is allow you to do color correction and allows you to pinpoint specific colors that need to be adjusted. So I'm using that as an analogy, to how an audio spectrum analyzer works. With a spetral analyzer you can see certain frequencies peaking out of the norm. So for the crickets, their chirps probably fall in a specific frequency band, if I had to guess, probably in the 2Khz-5Khz range. So by visually seeing the noise on the spectral analyzer you can narrow in on the center frequency of the cricket chirps, and then notch that frequency down. You can then narrow the bandwidth of that notch EQ and visually see how much the EQ notch is affecting other frequencies outside the chirping sounds and do a precision EQ adjustment, so you can turn down the cricket chirps, while minimally affecting the voices. The voice might not sound perfect, because the voice frequencies probably fall in the same frequency range, but you will end up with something where the cricket chirps aren't so predominant in the recording, and minimally affecting the voice as I mentioned. Sound Forge has a Spectral Analyzer in it, although this is not my favorite one but it still may be useful. Noise reduction won't work so well on cricket chirps as you noticed because it's really not a constant tone, because the crickets are randomly chirping at different times.
richardfrost wrote on 3/30/2005, 10:06 PM
Thanks all of you. Gr8Steve - I meant audacity for filtering and stuff when we're not using Vegas. Of course we use Vegas for voice recording.

farss and rednroll - we will try your tips tomorrow, thanks a lot guys.