Audio Mixing

[Will Hanlon]

Is there an instructional DVD for audio mixing in Vegas?

Going above 0db means clipping and the red indicators? So you can use the output fader to avoid clipping. The loudest sounds should be at 0db, correct? I'm just trying to get this all clear in my head.

At one point in my project, I have a gun shot that I want to be as loud as possible, but it never sounds that loud when I play it back on a TV even if it's in the red (if it's in the red does Vegas automatically lower it or something?).

What do you use buses for? Help in Vegas says "you could route all your drum tracks to a bus so you can adjust their levels together without changing their relative levels." So you could route all your dialogue to one bus so then you could just adjust that dialogue? Of course, that's too simple since audio is at different levels between takes. Anyone here use buses a lot and if so, what for?

Do you apply an equalizer to all your dialogue first? Can someone just give me like a sample guide of the steps they take when working on audio mixing?

Thanks. I realize there are a lot of questions here, but I'd really like to get a better grasp on audio mixing techniques.

[Jamie Hellmich]

Will,

For simple audio corrections, I usually normalize all clips on the track, then use a volume envelope to boost or cut audio in particular portions or particular wave peaks.

Increasing the timeline zoom helps view the peaks in the wave file.

I don't usually pay a whole lot of attention to the meter, and more on the sound preview quality.

You can check the track FX settings in the Track List by clicking on that button. You there have access to the Gate, Compressor and EQ settings.

Jamie

[Douglas Spotted Eagle]

You need to pay attention to the meters.
You cannot go "above" 0dB. That's the end of the line, the ceiling, whatever you want to call it. You can hit 0dB so hard that the signal becomes very nasty...or you can barely slam 0dB so only small, almost inaudible clipping/brickwalling occurs.
the key is to get your peaks on final output in the -.2 range.

Avoid normalizing if you can; record good levels.

use the WaveHammer on the master, set to a preset of "Master for 16bit" for final output. Better still, apply the iZotope Ozone filter to the master for maximum quality and awesome dithering quality.

My workflow?
Record solid signal levels
import
Sweeten after de-noising if necessary.
Apply light compression unless the voice requires heavy compression.
Apply iZotope Ozone to master.
I do slightly widen any non-dialog tracks to fit the voice into the groove slightly better. I'll often kill a tad of mid-range in any music track to allow the voice to more adequately fill in the gap as well.
Depending on the project I'll often pump 1dB of 80-120Hz into the mix as well, for a slightly more rounded bottom.

[Will Hanlon]

Thanks for the info guys.

Douglas, what would you say are good levels? Why would you avoid normalizing (does it make the audio too similar)?

Why do you shoot for -.2dB range (is that to leave room for an error or something)?

Do you need Pro Tools for the iZotope Ozone plugin?

Why do you apply light compression? What does that do?

How do you widen non-dialogue tracks? How do you kill the mid-range of music tracks? What is the mid-range (is that the range where the human voice primarily is)?

What do you mean for a slightly more rounded bottom? Does Vegas have a plugin to just pump in 1dB of 80-120Hz?

I apologize for all the questions. I'm just hoping to learn and demystify the process. Thanks.

[Seth Bloombaum]

Originally Posted by Will Hanlon

Douglas, what would you say are good levels? Why would you avoid normalizing (does it make the audio too similar)?

Good field recording levels peak at between -12 and -6db. This is to leave headroom for an unexpected peak, or, for a transient peak that your meters don't show you. With good levels recorded, normalizing is usually not needed. "Similarity" has nothing to do with it. As Douglas mentioned above, use envelopes to mix a single track or multiple tracks over time. There are several methods for drawing the envelopes...

Originally Posted by Will Hanlon

Why do you shoot for -.2dB range (is that to leave room for an error or something)?

Yeah, leave a little room. Different engineers will peak their mix anywhere from -1.4 to -.1db. This is not broadcast standard, which is a different ballgame.

Originally Posted by Will Hanlon

Do you need Pro Tools for the iZotope Ozone plugin?

No. Ozone is an excellent Direct-X mastering plugin that will work within Vegas, Sound Forge, and other programs that use Direct-X plugins. There is also some iZotope code included in Ultimate-S from vasst.com that has the functionality you'd want for mixing most video tracks.

Originally Posted by Will Hanlon

Why do you apply light compression? What does that do?

Briefly, lightly compressing a voice track will narrow the dynamic range, in effect allowing you to turn up the lower volume portions without turning up the higher volume portions. You should read up on compression and volume maximization - this is key to finishing an audio track or mix.

Originally Posted by Will Hanlon

How do you widen non-dialogue tracks? How do you kill the mid-range of music tracks? What is the mid-range (is that the range where the human voice primarily is)?

"Widen" I'll leave to Douglas.

Killing the mid range involves using an equalizer to bring down the volume of the music in the frequency range that competes with the human voice, say, 800Hz to 5KHz, but especially about 1.2KHz to 3.5KHz, depending on the voice and the music. Having good reference monitors becomes pretty important for this kind of work... as well as developing your ear.

Originally Posted by Will Hanlon

What do you mean for a slightly more rounded bottom? Does Vegas have a plugin to just pump in 1dB of 80-120Hz?

Use one of the equalizers available in Vegas.

Originally Posted by Will Hanlon

I apologize for all the questions. I'm just hoping to learn and demystify the process. Thanks.

Forum contributors are happy to help. Having decades of experience I still ask questions and learn from these forums, there's always something new to learn about.

If you can get the basics of audio mixing and mastering it will serve you no matter what tools you are using... but Vegas is a great audio-for-video tool.

[Will Hanlon]

Originally Posted by Seth Bloombaum

Good field recording levels peak at between -12 and -6db. This is to leave headroom for an unexpected peak, or, for a transient peak that your meters don't show you.

What I did for my film since I used an HVX200 is I recorded each channel on different levels so if one peaked, my other would be alright, and I think that worked out fairly well.

Originally Posted by Seth Bloombaum

With good levels recorded, normalizing is usually not needed. "Similarity" has nothing to do with it.

What does normalizing do then? Why use it? Hmm I think I may have found the answer to this: "This process analyzes the audio in your clip and adjusts the volume to make the loudest portion full-scale or 100-percent volume. Normalizing audio clips ensures maximum volume from even the quietest take." So basically normalizing helps if you have a clip with levels too low/too quiet.

Originally Posted by Seth Bloombaum

As Douglas mentioned above, use envelopes to mix a single track or multiple tracks over time. There are several methods for drawing the envelopes...

What are some of these methods? How do you use envelopes to mix a track?

Originally Posted by Seth Bloombaum

Briefly, lightly compressing a voice track will narrow the dynamic range, in effect allowing you to turn up the lower volume portions without turning up the higher volume portions. You should read up on compression and volume maximization - this is key to finishing an audio track or mix.

I'll definitely do some reading on that. I thought you wanted a wide dynamic range though... I have a lot to learn. Any good threads, book, sites you can recommend for reading about compression and volume maximization?

Hmm... I found this interesting bit on a Videomaker page:
"Next on the list are two rounds of dynamics control. The first step is compression - minimizing the difference between the loud and soft passages of your video. Go easy on this: you have the power to completely destroy all dynamics. Great dynamic range (pianissimo and forte) is an important aspect of classical music and can be why it is so exciting. Conversely, a lack of dynamic range (loud, loud, loud) is one reason why rock music can be so monotonous. When using the Sound Forge Dynamics effect, for example, I like a threshold of -20dB, a ratio of 1.3:1, an attack time of 15mS and a release of 150mS. This is a good starting point, but listen to the results before moving to the next step. You may want more or less compression. The second step is limiting. It's identical, but uses different settings. This time use a threshold of -3dB, a ratio of 10:1 and attack and release times of 0.0mS. The resulting image in the waveform view should be big and fat - just what you want. For comparison sake, turn the effect on and off as you preview the audio. The new version should be even and more consistently loud with a minimum of variance between vocal clips."

But I'm not sure what that means by threshold, ratio, attack time/release time. And I haven't heard anything about limiting before this.

Originally Posted by Seth Bloombaum

Killing the mid range involves using an equalizer to bring down the volume of the music in the frequency range that competes with the human voice, say, 800Hz to 5KHz, but especially about 1.2KHz to 3.5KHz, depending on the voice and the music. Having good reference monitors becomes pretty important for this kind of work... as well as developing your ear.

What is a good reference monitor? Can you point me to one? How does one of those help you? Does it show you different frequences or something?

Originally Posted by Seth Bloombaum

Forum contributors are happy to help. Having decades of experience I still ask questions and learn from these forums, there's always something new to learn about.

Well, thank you. These forums are really a fantastic way of learning thanks to everyone contributing.

[Will Hanlon]

Sorry I just thought of another quick question. What if I have some audio that peaks for like a millisecond? I have this scene where this girl screams when she's startled and the sound in both channels peaks... should I rerecord that sound or is there something I can do to fix it? It's incredibly brief, but I don't want my sound to be bad.

Just found I have another take where the audio doesn't hit 0dB so I should prolly use that take especially if her scream is just as good. How can you detect clipping/brickwalling? In the other take, it only hits the ceiling for 2 frames, and I know it doesn't hit it very hard, but I don't like it hitting at all.

[Daniel J. Wojcik]

Originally Posted by Will Hanlon

How can you detect clipping/brickwalling?

Your waveforms will be flat on top.

If you're using meters, too, they'll be red. (And if you have SoundForge and open the audio up in that, the meters will say "Clip" up top.)

[Douglas Spotted Eagle]

Originally Posted by Will Hanlon

Sorry I just thought of another quick question. What if I have some audio that peaks for like a millisecond? I have this scene where this girl screams when she's startled and the sound in both channels peaks... should I rerecord that sound or is there something I can do to fix it? It's incredibly brief, but I don't want my sound to be bad.

Just found I have another take where the audio doesn't hit 0dB so I should prolly use that take especially if her scream is just as good. How can you detect clipping/brickwalling? In the other take, it only hits the ceiling for 2 frames, and I know it doesn't hit it very hard, but I don't like it hitting at all.

You might try the demo of the "Clipped Peak Restoration" tool from Sony. It will potentially fix the audible portion of your clipping.
Bear in mind, there are two means of normalizing. You can normalize by RMS, or normalize by peak. Vegas cannot normalize measuring RMS. Sound Forge does.

[Will Hanlon]

My question about clipping/brickwalling wasn't clear enough. I meant what does it sound like? I know the indicators in Vegas of when it happens, but what do you listen for in the actual audio that indicates clipping? I've noticed it in some indie films when it's really bad, but that's because it was really bad (and I thought I heard it at one point in Superman Returns when Kevin Spacey yells, "Wrong!" but I could be nuts).

[Bill Ravens]

There are several ways to skin the "mastering" cat. Douglas' methods are certainly good advice. I, personally, like to use a little "normalizing" in the workflow. Normalizing as applied by Vegas, is fairly useless and can be very damaging. I much prefer normalizing in soundforge, with the "apply compression if clipping occurs" feature turned on. you can set the floor of the normalizing to avoid raising the volume of the noise floor. you can also select the threshhold level you're going to normalize to. it works well. It can also be said that if you're going to do any audio mastering, you're better off doing the whole thing in a single application like ozone at 24 bit, dithering to 16bit for the final output. working at 16 bit and going back and forth to 24 bit will corrupt the final sound enough that you should avoid doing it. staying in a single app like ozone will keep the conversions minimized.

[Daniel J. Wojcik]

Originally Posted by Will Hanlon

My question about clipping/brickwalling wasn't clear enough. I meant what does it sound like?

Whenever I've done it, it has been with music, and that sounded like...um..."a loose wire", I guess would be a good way of describing it. A crackly, staticy thing.

Or, perhaps, like some just drove a screwdriver through the speaker, if it's a one-time peak hit.

[Seth Bloombaum]

Originally Posted by Will Hanlon

What does normalizing do then? Why use it? Hmm I think I may have found the answer to this: "This process analyzes the audio in your clip and adjusts the volume to make the loudest portion full-scale or 100-percent volume. Normalizing audio clips ensures maximum volume from even the quietest take." So basically normalizing helps if you have a clip with levels too low/too quiet.

Right. Normalizing can quickly adjust the volume of a clip to peak at, say, -.2db. However, if your original recordings require this treatment they may have other problems from low recording volume as well. Good field technique, applied in a few minutes, can save hours of post.

What are some of these methods? How do you use envelopes to mix a track?

Well, the simplest is inserting a volume envelope on a track, then double clicking on the envelope to create a series of points, then dragging the points up or down to increase or reduce volume. You'll quickly find that it takes 4 points to define a volume change and return to previous gain. You should probably start there in developing your chops with Vegas audio editing.

I thought you wanted a wide dynamic range though... I have a lot to learn. Any good threads, book, sites you can recommend for reading about compression and volume maximization?

This is where art comes in, not science. There is no rule... classical listeners with good playback in good rooms expect to hear a wide dynamic range on recordings. That recording would be useless in my car, all quiet passages will be lost to road noise. Pop music has an extremely narrow dynamic range and is well suited to listening in a noisy car, or work environment.

Take some example CDs and rip them in your computer, pull them into vegas, and watch the meters; you'll see. Your video mixes should probably come in somewhere between these two extremes. Track compression is applied so that (roughly) sfx and music don't mask dialog. Or maybe you want that solo flute run to be heard. You might draw a volume envelope to do so, or use a compressor.

Hmm... I found this interesting bit on a Videomaker page:
..."When using the Sound Forge Dynamics effect, for example, I like a threshold of -20dB, a ratio of 1.3:1, an attack time of 15mS and a release of 150mS. This is a good starting point, but listen to the results before moving to the next step. You may want more or less compression. The second step is limiting ..."

But I'm not sure what that means by threshold, ratio, attack time/release time. And I haven't heard anything about limiting before this.

Threshold etc. are the basic compression controls you will find on almost any hardware or software compressor, including the Vegas track compressor.

What is a good reference monitor? Can you point me to one? How does one of those help you? Does it show you different frequences or something?

Reference monitors are good speakers designed to play what is in the recording. Home stereo and computers speakers are designed to make any recording sound good. If you're making this stuff, you want the former.

Jay Rose (dplay.com) has some great books about sound for our dv/prosumer hardware and software (online and some bookstores). Ty Ford has some kind of short manual for field sound that people like (tyford.com), also check out the "Now Hear This" audio forum for his contributions.

I'm sure there are many more, check out the "Read about it" forum as well.

I could go on at much greater length about sound, so could several other forum contributors, but you need some book-length info and lots of practice and listening.

Here are some random links:
http://www.vasst.com
http://dplay.com
http://www.recordingwebsite.com/articles/eqprimer.php
http://www.kenstone.net/fcp_homepage...ion_sound.html
http://www.equipmentemporium.com/Lavarticle.htm
http://www.locationsound.com/cgi-bin...chtips8_4.html
http://shure.com/ProAudio/TechLibrar...cles/index.htm

[Will Hanlon]

I've read a couple of those websites. The one on location sound I referenced a long time ago before I shot my film. It definitely came in handy.

I just looked up some reference monitor speakers... M-Audio Studiophile BX-5A... anybody have specific brands they recommend? Anyone actually own some speakers like these?

So let me see what I understand here. Good field levels tend to peak between -12db and -6db. Then you want your final audio mix to peak between -1.3db and -.2db (is this a good range for a feature film?). Thus, mixing is about raising the volume and clarity... ok not necessarily raising, but volume is a large part of it, and it depends on what you want (volume maximization I guess is the better term). Compression helps with clarity, etc.

So you could use a volume envelope to raise the levels of your dialogue (or sound fx, whatever)... I suppose how loud depends on what you want. Do you want your dialogue to peak between -1.3db and -.2db in the final mix or is that just the final mix of all the sound together? Would that be yes if you wanted your dialogue to be the loudest at that point in the mix? Sorry this is a poorly worded question.

I would think you wouldn't want a lot of volume increases and reductions in one string of dialogue, right? Prolly just a flat increase or decrease since otherwise, the listener could hear the manipulated fluctuations in that dialogue?

I definitely think I'll get the ozone app... supposedly they had good presets too so maybe someone relatively new could get a grasp on this. And Sony Noise Reduction 2.0... for some reason, I figured that came with Vegas... don't know why. It doesn't come with Sound Forge, does it?

Bill, you talk about not wanting to go back and forth between 16bit and 24bit since it'll damage the sound. I see you can set your audio bit depth in Vegas... so I guess I'd want that set at 24bit and then use the Ozone plugin? Or would I want to leave Vegas at 16bit? The higher the bit depth, the better?

Also, about setting the floor of normalizing in Sound Forge so you don't raise the noise floor, how do you do that? Is that fairly simple or how do you know what the noise floor is?

And Douglas, you apply compression using Vegas before you apply Ozone to the master? I thought ozone had good compression tools that you'd want to use instead of the compressor in Vegas.

[Bill Ravens]

Originally Posted by Will Hanlon

I just looked up some reference monitor speakers... M-Audio Studiophile BX-5A... anybody have specific brands they recommend? Anyone actually own some speakers like these?



So let me see what I understand here. Good field levels tend to peak between -12db and -6db. Then you want your final audio mix to peak between -1.3db and -.2db (is this a good range for a feature film?). Thus, mixing is about raising the volume and clarity... ok not necessarily raising, but volume is a large part of it, and it depends on what you want (volume maximization I guess is the better term). Compression helps with clarity, etc.

So you could use a volume envelope to raise the levels of your dialogue (or sound fx, whatever)... I suppose how loud depends on what you want. Do you want your dialogue to peak between -1.3db and -.2db in the final mix or is that just the final mix of all the sound together? Would that be yes if you wanted your dialogue to be the loudest at that point in the mix? Sorry this is a poorly worded question.

I would think you wouldn't want a lot of volume increases and reductions in one string of dialogue, right? Prolly just a flat increase or decrease since otherwise, the listener could hear the manipulated fluctuations in that dialogue?

I definitely think I'll get the ozone app... supposedly they had good presets too so maybe someone relatively new could get a grasp on this. And Sony Noise Reduction 2.0... for some reason, I figured that came with Vegas... don't know why. It doesn't come with Sound Forge, does it?

Bill, you talk about not wanting to go back and forth between 16bit and 24bit since it'll damage the sound. I see you can set your audio bit depth in Vegas... so I guess I'd want that set at 24bit and then use the Ozone plugin? Or would I want to leave Vegas at 16bit? The higher the bit depth, the better?

Also, about setting the floor of normalizing in Sound Forge so you don't raise the noise floor, how do you do that? Is that fairly simple or how do you know what the noise floor is?

And Douglas, you apply compression using Vegas before you apply Ozone to the master? I thought ozone had good compression tools that you'd want to use instead of the compressor in Vegas.

I'm using some M-audio ref monitors. reference monitors help you to hear everything that's in your audio. keeping in mind that your audience may not be using good speakers, playing back your audio on cheap speakers is equally as valid. It's best to do both, actually. And always check a mono audio signal to be sure your stereo signal isn't causing weird cancellation effects.

The subject of compression/limiting is very ...ummm...political in nature. The process makes the overall audio sound louder by raising the lower volume part of the audio and leaving the peaks unchanged. It will raise vocal and musical sounds, if they are in the same frequency range, equally. Modern music technicians are all making the choice to raise the volume of their work as much as possible. This is a personal choice.

My comments, re: ozone...Ozone is a one stop shopping kind of tool. Within Ozone are EQ, compression, limiting, reverb, and phase tools. The highest quality audio will keep the working file in the highest bit space for processing, then dither out to 16 bit for final recording. If you use a bunch of different plugins for each of these functions, you keep going back and forth between 16 bit and 24 bit everytime a new function is called up. Upconverting from 16 bit to 24 bit is rather useless, but, dithering back down to 16 bit space each time introduces distortion everytime you do it. So, keeping everything in a common tool and downconverting to 16 bit one time, will result in the highest quality audio.

I think the undisputed king of audio plugins, at least in the recording industry, is WAVES, but, these plugins will cost you an arm and a leg. Arguably, unless you're doing professional audio, Ozone will work very well for you.

Vegas and Soundforge, both, will let you look at the waveform of the audio. If you playback the audio in a portion of the waveform where there is no music or vocals, and you watch the VU meter, you'll get a sense of what your noisefloor is, usually around -90dB. This is what I use to set the floor of the Soundforge normalize tool. Soundforge will actually tell you what that noise floor is by using a file INFO tool.

Hope all this helps.