| Page (1) of 1 - 07/29/07 |
 email article |
 print
page |
|
The latest release of Sony Sound Forge 9 brings with it the Sony Noise Reduction plug-in. This invaluable tool has been around for a while, but was always a separate purchase. Now if you buy or upgrade to version 9, the plug-in is included free.
The Sony Noise Reduction package is actually three plug-ins: Noise Reduction, Click and Crackle Remover, and Clipped Peak Restoration. All three are useful for fixing the many field audio mistakes that seem to come my way these days. It is the general noise reduction plug-in that is the focus of this tutorial, though.
While the plug-in can work wonders, it doesn?t work miracles. It is most effective for reducing steady noise on sources that already have a decent signal to noise ratio. If your dialog is buried in the mud, it?ll never come out squeaky clean.
Here?s a typical example: an interview with some A/C noise and general din in the background. It?s most noticeable during the gaps between speech phrases. Before resorting to the fancy plug-in, I often run a gentle high- and low-pass EQ filter on the sound at 100 Hz and 10,000 Hz respectively. Use the Process > EQ > Paragraphic EQ for this task. It clears up rumble in the low end and hiss in the top. Already the track starts to sound better. I may tweak other EQ bands based on circumstances, but this is a solid start.
The Sony Noise Reduction works by taking a sample of the noise. After using the EQ, make a tiny selection of the noise. Usually a small sample, say under one-third of a second is best. Be careful with what you choose. You don?t want to catch the end of a word, a breath, or part of an intermittent sound or other one-time noise. You want a representative sample of the background noise to eliminate. Also, press the ?Z? key after making your selection to make sure it snaps to the zero crossing. This avoids clicks/glitches in the sample selection.
Go to Tools > Noise Reduction to launch the Noise Reduction plug-in. If you are not using Sound Forge 9, but own the plug-in already, you?ll find it in FX Favorites. You may need to organize your FX favs to have access to the plug-in from the menu.
There are several controls associated with the plug-in. The most useful are Reduction Type, Reduce noise by (dB), and Noise bias. Reset the plug-in by choosing the Default for fast computers Preset. Set the Reduction type to Mode 3. Of the three modes, mode three does the least amount of noise reduction, but has fewer residual artifacts. Mode 0 is very aggressive with reducing noise, but you will notice more unusual swishing sound in the output. Feel free to tweak and test this setting with your program material.
Click the Capture noiseprint checkbox and then click Preview once. The plug-in plays the selection and captures the noise. Click Preview again and you?ll hear the noise reduction already working (but only on your sample area.)
Click the Noiseprint tab to see a frequency plot of the noise. Right-click in the workspace and choose Filled graph to see it more clearly. Here the noise is in green, noticeably more prominent in the lower frequencies than the higher ones. The plug-in actually divides up the material into many frequency bands and develops reduction based on that. Though you can tweak these bands manually, it is usually unnecessary.
Switch back to the General tab. Right-click in the blank area on the right of the dialog box, below the Real-time checkbox. Choose Select All Data from the pop-up. Now when you Preview you will hear the whole file play, not just the noiseprint selection.
It is perfectly acceptable to tweak the Reduce noise by (dB) and Noise bias sliders while you Preview. You will notice that as you increase the noise reduction, the speech will start to sound funny ? a metallic, swishy-swirly sound. You are doing too much noise reduction. Back off on the amount and/or adjust the Noise bias, typically to the left.
A better way to use the plug-in is to check the Keep residual output button. Now when you Preview you will hear only what is being reduced (not what?s left). If you hear speech with this feature on, it will translate into more artifacts on the output. Continue to adjust the settings for the best compromise between reducing noise and minimizing its impact on the program material.
It is always better to run the program a few times with smaller amounts of noise reduction than to try to get all the noise out with one pass. Take noiseprints from different areas in the file and treat each pass to 6-10 dB of reduction.
Once you are satisfied, uncheck the Keep residual output and then click OK to process the file. You will easily see the results of your hard work in the waveform display. After the noise reduction, you may run a noise gate on the file to further clean up the output. You couldn?t effectively do this before the noise reduction, though.
Often a little manual clean-up is the last step, such as eliminating a breath, one-time noise, or in this case, a lip smack.
Sometimes you need to be rather aggressive on the noise reduction to make it work, but the resulting artifacts ruin the program somewhat. If you can, disguise these artifacts with other soundtrack elements such as room tone/presence, background ambiences, and music.
Page: 1
Jeffrey P. Fisher is a Sony Vegas Certified Trainer and he co-hosts the Sony Acid, Sony Sound Forge, and Sony Vegas forums on Digital Media Net (www.dmnforums.com). For more information visit his Web site at www.jeffreypfisher.com or contact him at jpf@jeffreypfisher.com.
Related Sites: Digital Producer , Audio Video Producer , Digital Pro Sound , Presentation Master , Oceania , BN - Sony
Related Newsletter: DMN Newsletter , Loud Newsletter , KNews Newsletter , Digital Media Net , DMNForums , Tutorial Finder
Vegas Pro 8 + Free Vegas Seminar Series
AV-HS450 16+ Input HD/SD Switcher w/ dual screen MultiViewer
