You can use the Noise Reduction tool in Vegas just as well. Only difference is that it's a slightly slower setup, that's all. I do mention (or thought I did) that I was exporting to Forge, because I talk about exporting to Forge in that same section. I honestly don't have a copy of the DVD's to look at....Sorry if I mislead you in some manner. Either way, it works fine in Vegas. Just insert it as a plugin, define the noise, and run it on all data either destructively or non-destructively.

I think what JCG is refering to is that the noise reduction plug-in does not necessarily come with Sound Forge... it is an extra product by itself.

I bought a version of Sound Forge which happened to come with the Noise Reduction 2.0 plug-in (or rather I wanted to buy NR 2.0 and the deal I found included SF 7.0 for less than the price for NR2.0 alone from the Sony site).

Liam,

Where'd you find this deal?

xog

I got it from the Student Software Store... but I purchased it at the non-acedemic rate (it was still cheaper than I had seen NR2 elsewhere).

I got it for $244.95.

Speaking of noise reduction, this is an area of much interest to me.
From what I can work out all the current noise reduction techniques are basically notch filters. SFs NR and Audition both use spctrum analysis to determine the noise spectrum from a noise sample and use that to setup the notch filters. This works fine with many types of noise but fails on others, particularly when there are lots of harmonics in the same band as vital information.

I had a tape to try and clean up with extremely bad noise due to a faulty mike lead letting in switching spikes from a light dimmer. Trying to filter this out was hopeless. But what i could get to work on a small section was subtracting the actual noise waveform.

I took a sample of the pure noise, inverted it and put it on a new track. With careful matching I could get it to completely cancel out the noise from the voice with no loss of speech quality over about a 5 second segement. It was impossible to extend this as the noise drifts very slowly in fundamental frequency however someone should be able to write some code to make this a practical method of noise reduction for this type of noise. It certainly will not work for mechanical noise, so the it doesn't replace the current techniques.

I might add that the noise was at about -10dB, so bad as to make the speech almost inaudible in places.

I eventually got it good enough for the clients purposes using Eq.

Brilliant

Farss this is brilliant! This is something I've been thinking about for sometime. Do you have a simple tute for me to do this? Nothing fancy, just: Step 1; Step 2; Step 3 . . . etc etc . .

Now, here's another thought. I've been reading about your trials and tribs on the Canopus thingy. Could the light dimmer be affecting your Canopus - only a thought . . but that'sfor the other thread - yeah?

Grazie

Grazie,
boy did all that get a bit mixed up!

1) The light dimmer thingy was on a tape someone gave me sometime ago to try to 'fix up'. I'd seriously suggested they reshoot it but just didn't have the time. I was only playing around with the idea of adding an inverted signal to cancel noise, I was able to demonstrate that it could work for his type of noise that has a consistant wavform and failry stable period. It will not work for real noise which is by definition random.

I got the idea from a technique used many years ago in valve radios to get rid of hum, They added more hum in but out of phase so it cancelled out. It works because hum is not noise in a pure sense of the word. In the case I was talking about I had these constant spikes about every 10 mSecs.

If you want to have a bit of fun you could try this. Say you've got a mulitrack mix of some music, bass, lead, drums and vocals. In VV bring in the final mix on one track and say the bass on another. Invert the phase of the bass. Now if you can keep the bass track in sync with the bass in the mixed track you'll completely remove the bass. You'll need to adjust the levels and get everything exactly in sync. It works best at the low end because the phase is easier to keep steady.

2) I don't have the Canopus box (DAMN, DAMN, DAMN), I have the ADS Pyro A/V link and as I and a few others are finding out its a real dog. I just tried it with some generated media i.e. SoFo evolving luminous objects, rendered out as a PAL AVI, bought back into VV and then using the Pyro to drive the monitor, at random points I get about half second flashes of green. I'll try shielding it a bit better but as it seems others are having similar problems it doesn't seem to be unique to my setup / location.

Sorry - me bad!

Thanks for the filtering tip . . . that's a keeper!

G

>>I took a sample of the pure noise, inverted it and put it on a new track. With careful matching I could get it to completely cancel out the noise from the voice with no loss of speech quality over about a 5 second segement. <<

I wanted to try something similar. I recently did some recording in a crowd. I used a decent directional mic on one channel and left the camera's internal mic on the other channel. In some cases it is hard to hear the voice I am aimed at because of the background noise of the crowd.

I thought I might be able to improve things by inverting and adding a portion of the more omni channel to the directional channel. I haven't been able to try it yet because of two problems.

I recently purchased Acid Pro which came bundled with Sound Forge Studio 6 (the low end version). I tried to use it and it seems that even inversion is not in this product. Is this cheap Sound Forge good for anything? I have Vegas and Acid Pro. I can't think of any reason I need this stripped down version of SF.

So I have a version of Cool Edit Pro too and it should be able to do what I want but when I tried it the multichannel portion, where I planned to add the channels, seems to be broken now. Crashes even on simple tasks. It used to work. Guess I need to reinstall, but I haven't had time to try that yet.

Any comments on the usefulness (or not) of SF Studio will be appreciated. If I get a tool working for the 'inversion and add' I'll try to post my results with subtracting one mic channel from the other.

Yep, nice trick there Farss!

While the noise reduction filter from SoFo is good, no single filtering method can get rid of every noise and no one application can do everything best.

Just before I got this damn Bell's Palsy (4+ months and still counting) I was all set to write some tutorials about the audio side of things. In doing research, I came across Arboretum's Ionizer. Which to give you an idea how much control you have it comes with a nearly 100 page manual (PDF) that deals with noise removal exclusively.

It works as a FX plug-in (supported by Sound Forge and Cool Edit, but I also got it to work direcly with Vegas) while I did try it out, I didn't really get that far into it because at the time I couldn't hardly see at all because of the BP thing.

The cool thing is you don't get the bird chirping or hollow artifacts anywhere near as much as you can get if you don't do that great a job building the notches in the filtering of the noise you can with some other products. It ain't cheap, $500, just mentioned so you know there are tools out there for tougher problems.

http://arboretum.com/products/plugins/ionizer_main.html

This is actually a reply to the posting from Farss, several messages up in this chain.
I think SoFo's Noise Reduction package, in one of its modes, actually works just the way you're describing -- you capture a "noiseprint" based on a noise-only part of your file, and then apply that to the file as a whole.
I used to use that quite a bit, back in earlier days---but ever since SoundForge starting using the "plug-in manager" setup, I can't figure out how to use the noise-cleanup function anymore.

I've been meaning to ask SoFo (oops--Sony) support about this, so this'll be a spur to finally ask. I'll post again, if I learn anything interesting.
Ernie

I'm kind of groping in the dark here... I only used a demo version of SoFo's noise reduction way back when and I no longer have it or the Inoizer installed.

If I remember correctly you capture a "noise" then you get to push little sliders up and down the graph to pinch it and shape it to get rid of the noise. In other words there's the waveform and a single envelope to adjust.

With the Inoizer they add another envelope that's EQ, based on bands. Its this additional envelope that better controls gain in tiny steps that makes all the difference in getting rid of tough noise.

earthrisers -

You don't have to run Noise Reduction from the plug-in chainer--use the DirectX Favorites instead. Then, the plugin is applied just as in older versions of Forge.

So, just select a passage containing noise only, select SF Noise Reduction from the DirectX Favorites menu, capture a noiseprint, and you're ready to go.

Of course, this method doesn't take advantage of Forge's non-destructive editing, but that shouldn't be a problem if you're being careful.

Not a direct Vegas question or answer, but what the heck.

RexA,
givn the separation between the two mics I doubt you'll have too much success. Noise cancelling mics have two mics back to back a few millimeters apart.
But you can do all this in VV, it has a track invert switch.