Newbie - Needs help improving audio quality

ArtVandelay wrote on 5/23/2005, 7:06 AM
Hi,

I have been using Vegas 4 for about a year, doing pretty basic editing. My wife "volunteered" me to put together a video yearbook for my son's kindergarten class. My wife shot the video and the audio quality is poor on many segments.

The finished video is to consist of short interviews with the kids talking about what they liked about kindergarten, their favorite thing in kindergarten etc.... On many of the segments the kids are talking so softly that the sound is basically inaudiable.

I can year the kids thru my pc with headphones on. I made a svcd for testing purposes and when I play it on my dvd player the volume is very faint on some interviews.

I added an audio envelope and raised the volume level during the interviews on the timeline, but the volume is still faint. Another problem is that there is a lot of background noise; either from other kids in the classroom or some sort of electric motor like an a/c unit.

I have to have the video finished by this Wednesday and burn 20 copies. Reshooting isn't an option, and I don't have the $ to buy an expensive plugin. Are there any tools contained in Vegas 4 that I can use to improve the audio quality?

I did buy Spot's Vegas 4 book and noticed there is a sample veg file on the noise reduction plugin, but I didn't know if that would work for my purposes.

Thanks

Comments

BillyBoy wrote on 5/23/2005, 8:13 AM
If you haven't tried it already, right click on your audio track, switches, Normalize. This will increase the gain, but also increase the noise. Removing noise is a iffy thing. Sometimes its easy, other times next to impossible. If the noise is in the background while the kids are talking because the "noise" shares some of the same frequencies trying to remove all of it may reduce the quality of what the kids are saying that doing some "correction" actually makes things worse.

Long ago I wrote a tutorial on removing camera noise. Click on my name to visit my tutorial site. The principle is the same.

While Sony's Noise Reduction filter is top notch, it is expensive, especially for casual use. There are many other audio tools, several priced much lower. One I like is DC Millennium that uses noise filtering similar to how Sony's pricy version does it. Audio Cleaning Lab is another that sometimes does well. The same company that developed Millennium now has a web site that allows you to upload a problem audio file and their engineers will run it through some high end stuff and probably make it much better. Do a Goggle search for Diamond Cutter.

Also they have a newer audio forensics tool that may be what you're looking for. They claim that several intelligence agencies use it like NSA, CIA, FBI, so who knows. maybe worth a shot.

http://www.diamondcut.com/Catalog/LiveForensics.htm


http://www.enhancedaudio.com/millennium.htm
zdogg wrote on 5/23/2005, 10:21 PM
There are a few of things you can try:

First of all, you might want to make copies of your audio tracks, and thus if your editing goes amis, you can easily get back.

1. Cut low and high frequencies, as voices are midrange. This will dampen some of the noise. Maybe try cutting or rolling off below 200 - 300 hz and above 5K.hz Maybe add a spiked boost, very narrow band at each 7k and 9 k to boost silibants for a little crispness and speech intelligiblity. You can do this before "normalizing." Also, normaliziing adjusts the gain of the whole track just to the point of the loudest section spiking. If your loudest section i say -1db, your overall gain will be just 1 db for the track. You can improve on this scenario by boosting the wave gain in sections, not as a whole track as does normalizing. IOW a real soft section you may be able to boost by 20 db, and again, the normalizng only gave you a 1db boost, IOW nomalizng boosts loud and soft sections the same, 1db in the above scenario.

2. cut the volume around the speech using your audio envelope, but be judicious here as you don't want to have it become so unnatural as when everything comes up together, noise and foerground, but by executing #1., this should be less of a factor.

3. Try using an expander, which works in the opposite way as a compressor. You expand louder foreground signals theoretically increasing your signal to noise ratio, but this sort of depends on how loud your noise is compared to the foregeound.. You will have to experiment with threshholds and expansion ratios in so doing.

4. Soft undertrack of music perhaps, to make the noise less standing out.

Z
Spot|DSE wrote on 5/23/2005, 10:31 PM
By chance, are you sure you have both audio channels going? If the kids are being interviewed, or talking, then you should have some audio there, especially if there is an airconditioner that's louder.

Next, I'd try an expander, coupled with compression. you might add a noise gate with slow release. Duplicate the audio tracks you have, and run one heavily processed while mixing it with the track that isn't processed. Try increasing high mids in the processed track. Listened to by itself, it will sound horrible. Mixed with the other, it should make a nice blend that will get you past the mess.
Cut anything below 100 hz. Use pad or ambience to fill the holes.
Noise Reduction in a room of noisy kids will be more or less useless. The airconditioner can be reduced if it's really that loud, you might try Sound Soap, for 99.00. I think their demo is 14 days.
Try the iZotope mastering tool, www.iZotope.com and see if that helps, there are some great "noise oriented" presets in there.
riredale wrote on 5/24/2005, 11:53 AM
If the volume is still faint after raising it on the timeline, then you can easily do this: raise the volume as far as you can, then render that audio track as a new wav file. Then bring that new wav file back into the timeline and use it to replace the original audio track. If you double-clicked on the original audio track and selected "render only the loop region" then the new wav file should exactly replace the original wav file. Once the new wav file is in place, you can adjust the volume on that file and raise the level even higher.

A final thought: if the audio on your DVD is much lower than what you had on your timeline, you need to adjust the "Dialog Normalization" setting which is used when rendering the audio into an AC-3 file.
riredale wrote on 5/24/2005, 11:55 AM
If the volume is still faint after raising it on the timeline, then you can easily do this: raise the volume as far as you can, then render that audio track as a new wav file. Then bring that new wav file back into the timeline and use it to replace the original audio track. If you double-clicked on the original audio track and selected "render only the loop region" then the new wav file should exactly replace the original wav file. Once the new wav file is in place, you can adjust the volume on that file and raise the level even higher.

A final thought: if the audio on your DVD is much lower than what you had on your timeline, you need to adjust the "Dialog Normalization" setting which is used when rendering the audio into an AC-3 file. I think the default is -27; if you change it to -31 then the resulting output will match the original. I don't know much about the DVD-A product since I use something else to author DVDs, but I think this is accurate info.
jlafferty wrote on 5/24/2005, 8:18 PM
Just wanted to say thanks to everyone contributing with audio tip, tricks and info. Audio seems to me the most precise, demanding aspect of post and conversely also the aspect most shrouded in mistery and obfuscation. Thanks for helping to clear the smoke a little.

- jim