Listening to the crowd sample there's not much that can be done. You might get some improvement using simple Eq to cut anything outside the speech band, you need to find the best balance between damaging what you want to keep and what you want to get rid of. I'd certainly cut out as much of the low frequencies as possible. That probably didn't sound too bad back when it was shot. The TV's back then only had tiny speakers. With today home cinema systems all that rumble from the crowd would sound pretty bad so I'd try to get rid of it. Just keep in mind the viewer will be aware it's vintage footage so don't panic over it not being perfect.

Bob.

I agree that this is not what Noise Reduction is made for, although it might help a little with the traffic noise. I also agree with Bob on the EQ treatment. Try that on what's left of the traffic after you NR'd it.
General tip on NR: Don't try to get all in one go. Do it several times, and cut only 8 dB each time. Important: Remember to take a new noiseprint each time!
I'd use the paragraphic Equalizer and experiment with reducing different parts of the spectre. The voice would probably be mostly at or around 1K, so leave that at zero.

Finally: Audio and Noise Reduction questions are not considered OT in this forum.

Tor

Hi,

this is a job for a special FX. I have done a lot of this, using Sony Sound Forge and two plug-ins:
- Voxengo Redunoise
- iZotope RX
(Search the web)

Sometimes Voxengo is better, sometimes iZotope. An equalizer is not the right tool for this. I have tried one of your samples, and managed to get the noise down. But I am in a hurry, so I cannot help you right now. It takes some time to learn how to use these plug-ins, the result, however, is amazing.
They are intelligent, and "learn" what the noise is, and then filter it out. All you need is a section with only noise, no speech, to "teach" the plug-in.

I always use these tools when narrating videos. Somehow noise will creep in, computer fans, system hiss etc. Not to mention fan noise from studio lights. For sound where background noise is several dB below voice, the result is absoultely brilliant. As a matter of fact, I use them as the rule, not the exception, when I want a crisp clear voice with zero background noise.
An EQ is a stone age tool compared with the ones I mentioned above. Although a combination often helps.

Edited:
Audacity (Search the web) is a freeware audio NLE, this has a plug-in that does this, although not as sofisticated as the two I mention above.

ingvarai

"You might get some improvement using simple Eq to cut anything outside the speech band, you need to find the best balance between damaging what you want to keep and what you want to get rid of. I'd certainly cut out as much of the low frequencies as possible."

Please forgive this Noob question, but what is "simple Eq" and what is the "speech band"?

...Jerry

"this is a job for a special FX. I have done a lot of this, using Sony Sound Forge and two plug-ins:

Thanks for the suggestion, but this is out of my price range for a freebee fix of a old 20 min video.

I might give Audacity a shot - does it really do more than Sound Forge? Also, I have Nero WaveEditor - would this be of any help?

...Jerry

I might give Audacity a shot - does it really do more than Sound Forge?

Audacity is a NLE (Non Linear Editor). Sound Forge is a "sound lab". Typically, you mix tracks in an NLE, then take a single sound clip into Sound Forge, give it the treatment you want, save it and send it back to the NLE.

Let us say you are mixing a song, and have guitar on one track, drums on another, organ on the third and song on the fourth. You use an NLE editor for this. Then, you want to remove background noise from the song track, and you import it into Sound Forge, do your stuff, and send it back.

In your case, what you need is something like Sound Forge. But - Audacity is capable of dping what you want, as a bonus. You get what you pay for. It is free, but the noise cleansing is fairly good.
Sound Forge is a professional tool, you cannot compare it directly with the free Audacity NLE

I have Nero WaveEditor - would this be of any help?
Never heard of it.
Good luck!

ingvarai

ingvarai,

"In your case, what you need is something like Sound Forge."

My apologies, I may have confused you... I have Sound Forge w/ the Noise Reduction plug-in. I tried the technique of taking small samples of the noise and applying the Noise Reduction several times - resulting in some improvement, but nothing dramatic.

I was hoping that the "experts" in this forum help me work some magic!

...Jerry

> was hoping that the "experts" in this forum help me work some magic!

Yes, the answer in your case is: Download the trial version of Voxengo Redunoise. This is the magic in person :-)

ingvarai

Jerry - in my opinion, the noise level is too close to the voice level to minimize it without noticeably impacting the voice.

Noise reduction plugins work best when the noise is constant (wind is tough to cut in post), and/or noticeably lower than the voice. Otherwise they will only slightly reduce the frequency range of the noise but also noticeably impact voice clarity, and if pushed too far, introduce phasing and harmonic distortions and reductions that can be less tolerable than background noise, esp. in a piece like this.

Redunoise is as good an option in standard plugins for doing something as any though.

For a piece with mostly personal value for people involved, I would go easy on the noise reduction - it can do more harm than good.

Regards,
Dedric

"I was hoping that the "experts" in this forum help me work some magic!"

This thread in the Vegas Audio forum was entitled, "Help a Novice fix the impossible." The experts in this forum have confirmed that it is essentially impossible. Audio processing software can do a lot of amazing things, just not this.

Despite the environmental noise on the tracks, the voices can be understood. I assume that the video that goes with this audio shows the crowd of people which would explain to the viewer why that tracks sound the way they do. Any further processing would surely make the audio tracks sound unnatural.

You'll find a perfectly adequate equalizer in the audio track header in Vegas. With that you can cut or boost a range of frequencies.
The spectrum of human speech is quite wide, from around 100Hz to 10KHz, some would probably argue even wider than that. You don't need all of it for it to be intelligble and it may be more intelligble if you cut some of it out. Some general guidelines here:
http://en.wikipedia.org/wiki/Audio_frequency

Note that speech which is simply "intelligble" is not pleasant to listen to.

I'd suggest you invest in a copy of Jay Rose's excellent book "Audio Postproduction for Digital Video", ISBN 1-57820-116-0.

Bob.