Basic audio touchups through Vegas/SoundForge

[Phil Gosselin]

Hi all,

I am by no means an audio expert, the client doesn't actually have the budget for a proper audio post production so I guess I will have to make it.

My shoot interviews in a convention with 3 subjects, each with a wireless mic. Sound is pretty good but I am sure that a few basic touchups might make it even better.

I am editing in Vegas so all the audio plugins are the same as in Sound Forge.

Mainly there are two things I would like to do:

1: Attenuate the background noise
2: Make the voice even cleaner

Which plugin should I use to make these two adjustments?

If you think of other plugins that might come in handy please suggest it, keep in mind though that a shot description of what that plugin does will be appreciated as I do not speak the audio "language".

Thanks,

Phil

[Steve House]

If the background niose is something constant and relatively narrow band like the hum of a fan you might be able reduce it with the noise reduction plugin but if it's random, like conversations or traffic noises, there's really not very much you can do about it. As for making the voice "cleaner," what you can do, if anything, depends on precisely what is wrong with it to begin with. There's no substitute in post for proper choice of location, proper mic placement techniques, and proper recording techniques on location when making the original master recording..

[Greg Miller]

Hi Phil,

What Steve said about noise reduction is spot on (as usual).

You probably wouldn't ask your physician to diagnose a "pain in the gut" over the phone, he would want to examine the patient. Likewise, if you want recommendations about cleaning up your specific track, you should post a few clips so we can examine the patient. You will get much better suggestions that way.

[Rick Reineke]

To elaborate on what Steve and Greg stated, The SCS noise reduction plug-in works well for attenuating 'din' type noise like HVAC, distant traffic, ground hum and EMI noise, ect. If you can post an audio clip, we could better suggest what can (and cannot) be done. What version of SF and Vegas do you have? SF-AS for instance lacks the NR-2.0 noise reduction plug-in package as well as others. Depending on the Vegas version, most of the SF plug-ins can be used in that as well.

[Chris Barcellos]

Since we are the subject here, back about half a year ago or more, there was a promotional Sound Soap given out for a short period. I have been running it in Vegas, and wonder what you sound gurus think about it. Also, in Audacity, there is a noise reduction filter that samples from a portion of the clip you select, and I wonder how these are received.

[Phil Gosselin]

Hi,

Well we shot the footage in a convention so the background noise is constant. Take a look at the screenshot attached, it shows a list of all plugins available with my version of Vegas, which is 10a.

Greg: I fully get what you are saying but I am only looking for a way to slightly adjust the sound with the minimum of effort. If it can be done then fine and if not I will just leave it as is.

A great parallel could be if someone wants to reduce red in his picture. In Vegas there is a "color balance" plugin that would allow to do that very fast. It is a very quick (and dirty) way of reducing reds, by far not the best solution but it will get the job done.

So if there is a quick way to slightly reduce background noise I will do it and most likely bring the overall quality of the sound up a notch. Which is more than enough and will make me happy.

Don't get me wrong, the sound is very good it's just that I am simply looking for a quick way to improve if possible.

Thanks gents, your time is greatly appreciated.

Phil

[Battle Vaughan]

Originally Posted by Chris Barcellos

Since we are the subject here, back about half a year ago or more, there was a promotional Sound Soap given out for a short period. I have been running it in Vegas, and wonder what you sound gurus think about it. Also, in Audacity, there is a noise reduction filter that samples from a portion of the clip you select, and I wonder how these are received.

My two cents, I like Sound Soap (have only used the basic version, but found it to be quite effective on narrow-band noise.) Likewise the Audacity noise filter, which works pretty much the same as all the sampling filters (Soundbooth, Sound Forge, etc) by eliminating whatever you can sample from a noise-only portion of the clip, works fine if you have specific, narrow-band noise (60cycle hum, etc.) Where all these plugins have a problem is wide-band noise that overlaps the part of the audio you want to save, as it affects the good stuff along with the bad. I have had rather surprising luck with Sound Soap, in one case almost eliminating a distracting din of several people talking in the background of my interview while enhancing the voice of the interviewee. The mac version works a treat, the PC version (IME) has/had a seriously flawed license verification scheme which made it almost useless. Maybe they have fixed this by now.

[Seth Bloombaum]

Still, there is no magic "quality" knob to grab and turn up.

In general, there are 3 classes of audio tools to consider:

EQ - this is the most basic, and often overlooked tool. Roll off the low end until the desired audio is affected, then back it off a little bit.

Dynamics - this includes volume tools and compression/expansion tools. I see Wave Hammer in the plugins list. It's a pretty good compressor. Compress to keep voices natural, but more intelligible, but watch out for the tendency of a compressor to bring up background noise. Clean recordings can be compressed more without objectionable background noise "pumping".

Noise reduction - the most basic is a noise gate, that tends to lower the volume when overall signal is below a certain threshold. Usually overlooked in video work, because it's only occasionally useful. Noise Reduction doesn't appear in your plugins list, but it can be pretty magic in its ability to reduce many unwanted noise sources, but not all.

If you've decided NR is needed, but you don't have such a plugin, Audacity (sourceforge.net) is an open source audio tool that includes a basic NR.

That's a basic orientation to the tools, but, to repeat what's been written above, none of us have heard your sound, so we can't make meaningful recommendations. Don't know whether you need a saw or a scalpel..

[Jon Fairhurst]

Noise reduction is a destructive approach, well described above. The challenge is to destroy the background sounds without overly destroying the dialog. Sometimes the background noise is bad enough that distortion of the dialog is a good trade off. In other situations, it's best to live with the background noise and focus on making the dialog crisp and clear. That leads us to...

The constructive approach. Here we add music and/or sound design to cover up the background noise and make it more pleasing to the ear. We can also EQ the audio to make the dialog as crisp, clear, and intelligible as possible.

When using music to cover the background, compress the music track so it can be played at a low level yet be consistently heard. You don't want it to come and go with the dynamics of the music. You can "duck" the music a bit when there is dialog, but be subtle so it's not really noticeable. If you keep the music low, you don't really need to duck it. You can also add a low-end sweetener or rumble. Watch a Star Trek film/episode for an example. If it's subtle enough, no one will notice.

Regarding EQ, I use the following as a guide. Use smooth hills and valleys, rather than sharp peaks for this style of EQ:

200-300 Hz: This is the voice fundamental. Adjust this so the voice is neither thin nor boomy. Young women have fundamentals at higher frequencies than old men. Adjust the center frequency as needed. Pull this down for music or sound design. You can boost 100 Hz for music if it needs more body.

600 Hz: This is generally unimportant for voice. You can pull this down on the dialog track and boost it on the music or sound design tracks.

1200Hz: This is critical for consonants and intelligibility. Boost on voice and cut for all else.

2400Hz: This frequency space is for nasal tones. It helps differentiate the voices of two different people or the sound of an oboe vs. a flute. If the speaker sounds harsh, tamp this down. If they sound dull, boost it up. If two people have similar voices, boost this to help differentiate them. If you can't tell the electric guitar from the cellos, boost for music. This is your call.

5000-10000Hz range: This is for "air". Boost to add breath and sparkle. Watch out for noise and sibilance. You can add this to both voice and music as you would add salt to a meal.

Resonance: If there is an obnoxious tone in the room, the voice, or music, you can find it by sliding around sharp peak with your EQ controls. Once you find the problem tone, remove it by creating a valley at that frequency. You can do sharp, narrow cuts with EQ, but never do sharp boosts. Any boosts should be smooth to avoid adding resonant peaks.

If you're looking to improve already good sound, rather than fix a poor recording, consider the constructive approach. Do it well and the voice will be clear and sweet and the audience will quickly forget about the sounds in the background.

[Phil Gosselin]

Jon and Seth:

This is truly gold advice, thank you very much. Indeed I've copied it in a text file for reference in future edits.

If it were up to me I would just pack all the audio and throw it to one of my freelance sound engineer and not bother at all. Unfortunately more often than not clients believe it is a waste of money to pay extra for audio post-production so I get stuck with it.

Thanks again !

Phil

[Steve House]

Originally Posted by Phil Gosselin

Hi,

Well we shot the footage in a convention so the background noise is constant. ...
So if there is a quick way to slightly reduce background noise I will do it and most likely bring the overall quality of the sound up a notch. Which is more than enough and will make me happy.

Don't get me wrong, the sound is very good it's just that I am simply looking for a quick way to improve if possible.

Thanks gents, your time is greatly appreciated.

Phil

The background sounds at a convention are highly UNLIKELY to be constant in tone and quality (compared, say, to the constant drone of an AC blower motor in the background). They are constant in that they are always present, true enough, but the frequency spectrum they present will be continuously changing all over the place.

[Greg Miller]

Originally Posted by Phil Gosselin

A great parallel could be if someone wants to reduce red in his picture. In Vegas there is a "color balance" plugin

With all due respect, since you are asking for advice, and admittedly don't have a lifetime of audio experience, I have to disagree with your analogy.

With video, there are three primary colors. We all know what they are. Less red means less red. Imagine someone telling a chef "this dish needs less flavor." The chef needs to know which flavor is the problem: is there too much salt, too much pepper, too much paprika, too much cumin, etc. Noise is like that: there are more than three "colors" of noise, and different noises can overlap, so you need different approaches to clean up the track.

Experienced audio folks describe many different types of noise. Single or multiple steady tones, hum or buzz (two vastly different things), hiss, rumble, random noise, high frequency noise, low frequency noise, etc. Beyond that, some people unacquainted with the jargon might hear clipping or distortion and call it "noise."

Unlike a single "color balance" plugin, there are many different plugins or tools that are used to reduce the different types of noise. Different tools for different noise.

You say "the background noise is constant" but... what type of constant noise is it? Someone with more experience might make a more specific assessment... might hear a few specific constant frequencies (e.g. from HVAC motors), or a constant low frequency room rumble, or just constant random crowd noise. Having made that expert assessment, he might be able to give you more specific (yet simple) ways to reduce it.

Indeed, you have gotten some good basic audio advice in posts above this one! You won't go wrong if you digest, absorb, and use that information. But it seems to me you still haven't gotten a good specific answer about reducing the noise... simply because we haven't heard it yet. This isn't a criticism or a complaint, simply an offer to provide more help if you provide more [audible] data.

And, by the way, keep this in mind: any EQ you add to improve the quality of the desired voices will likewise effect all the background noise.

[Jon Fairhurst]

Steve, that's true, but partially so. The drone at a convention is made up of many, individual voices that change and are unique, but with enough voices, it becomes a bit more like random noise. And the mic and hall will give it an overall frequency footprint. You won't be able to remove the voice of the loud guy who walks past, but you might be able to reduce the underlying drone.

One problem is that the drone isn't far from the response of the dialog. Push it too far and the dialog will get that hollow, underwater sound.

When I first used noise reduction, I pushed it too far. My recommendation to the novice is to make it sound good to the ear, then back off the noise reduction to the point that you roughly double the noise. I think the problem is that the editor focuses on the noise, so they become sensitized to it. The casual user won't care about the noise, so long as things don't sound unnatural and can be easily understood.

[Jim Andrada]

I've sometimes been able to make things sound better by actually ADDING noise - done judiciously it can mask some low level problems. Sort of like the intentional motion blur that CD applications add to give smoother appearing videos.

On definition of noise is - stuff you don't want the listener distracted by.

[Greg Miller]

Originally Posted by Jim Andrada

One definition of noise is - stuff you don't want the listener distracted by.

So he should use the Distraction Filter to clean it up?