How to Produce Voice Over's that Don't Suck Part IV: Editing and Post Production

How to Record Voice Over's that don't Suck, Part IV: Editing and Post Production

Yes, I'm finally back from a looooong hiatus in posting blogs to GG. I hope to be more regular from now on in my contributions. Many exciting projects got in the way!

At any rate, if you have forgotten what has happened before, please see the bottom of this post for older blogs that lead up to this one.

Before we get started though, I'd like to write a small disclaimer. Audio engineering is far more art than science and the way that I present some of my "solutions" here will not work for everyone nor for every situation. There is a great deal of experimentation involved for both the novice and professional alike. Also, technology differences (different software, hardware, etc.) in your setup will affect the choices and tools you use. Finally, I'm assuming some basic knowledge of how these programs work and instead going to focus on sort of a "tips and tricks" approach. That being said -- let's get started!

Here we go! -- Volume Automation

At this point, we've assumed that you've recorded some VO (Voice Over) and at this point just need to massage the sound a bit as to make it more pleasing to the ear. For your entertainment, I've included part of a voice over from Wildlife Tycoon: Venture Africa ---

No fx applied MP3 sample

Not "bad", but it could use a little work. In fact, it actually had a little work done as far as levels. Are concerned before I bounced this clip for you to hear. Chiefly, I used some volume automation to lower the harshness of some of the constants that this actor performed. A lot of audio engineering is taking a great performance that has just a few issues and "saving it". This was one of those times. You can see in the screen shot below some of the automation that was inserted. (The yellow line across the large waveform).



This automation, which you draw with the mouse or other controller, then has a direct effect on that particular parameter in real-time. Say for instance that you wanted to mute a track because the actor coughed or breathed loudly between phrases... simply enact automation for a "mute" function and voila! Muted in exactly the right spot. Most DAW's (Digital Audio Workstation like Logic Pro (shown here) Cubase, Pro Tools, Digital Performer, etc.) worth their salt have such a feature.

It should also be worth pointing out from the image above, that the way the entire session was constructed was purposeful. There were around 25 VO's in WTVA that all had similar background fx and music. So, to construct the session, I created a track for each VO that was going to be recorded and then laid in the background fx and music on seperate tracks. In this way, without opening and closing a new project every time, I could record one VO after another. When it comes time to bounce the track to a stereo file, I simply "solo" the music/fx and the VO I'm bouncing and the rest of the VO is muted.

First Pass -- EQ

For EQ, I had a few simple goals. One, I wanted to make the actor's voice sound a bit warmer. Second, I wanted to make sure that it clearly cut through the rest of the mix. To accomplish this, I turned to the fairly advanced Channel EQ plug-in that comes with Logic Pro (I didn't use any 3rd-party plugs for this demo, only those included with Logic Pro, though there are many others out there for lots of different DAW's that achieve similar results). I captured this screen in real-time to show how it can be used.



The blue line is the real-time wave form. It moves along with the audio to show you what frequencies are being produced more than others. The green shape is sculpted by you and indicates where in the frequency band you are affecting the sound. Low to High is left to right and louder to softer is top to bottom. Around 1K is where a lot of human speech takes place and is the "clearest" part of speech for many. For this reason I have it a slight boost of around 2db at this range. Notice that I made a nice slope though as well. You can make this very pointed, but the sound turns very harsh. Then, I also gave a slight increase of again around 2db at 200 Hz. For this man's voice, that seemed to really warm it up a bunch. However, at about 120hz, I quickly dove below "0" to make sure it didn't become "boomy" and overly "bassey".

EQ is a lot of fun to play with. Most newbies at it apply way too much because they see how much CAN be applied and really distort the sound. For most cases, less than 5 db in either direction does the trick -- especially for voice.

Next pass -- compression

Now that you are dealing with the right frequency band, time to gently compress the signal. Compression is one of the most useful (and used) plugs/effects used in modern music production. However, it has been in used in one form another by the pros for decades! The most important part of this control is the ratio. Here, I'm compressing 2.2:1.



I couldn't tell you the exact math on how that works, but the higher the first number, the more compression that is applied. Basically, it limits the dynamic range possible in the signal by bringing up some of the quieter stuff and bringing down the loud stuff. Sounds simple, but if used improperly, this can radically alter your sound. In general though, a little lite compression gives it that "radio sound" (which is highly compressed) and evens things out tremendously. Yes, it will also "seem" louder (and will be) but this is mostly because the dynamic band has been limited.

Reverb




I didn't use a lot of reverb on this track, but just a little for space can really help make it sound like the voice wasn't recorded in a closet. (Which it basically was.) Notice how low the "reverb" slider is set for here. Also, the preset called "Clear Vocal" is a very short reverb (.690s) anyway, so this is not a "hall" type sound or anything. It's simply meant to enhance the sound.

Master Track

Finally, I'll speak a little bit to your master fader. The master fader is the last one the signals pass through before they go to your speakers. Essentially, all the audio gets processed here, so make sure that whatever plugs you use work for everything! There are two that I use fairly consistently.



The low cut simply cuts out frequencies lower than the value you set. However, because of various technicalities, this *can* increase the overall dynamic just a hair in the remaining registers. I usually set this for 64-72hz. Why? Granted, most of us can hear or at least feel down to around 20-25hz. However, most *speakers* don't actually reproduce sound that low because of the physics in speaker construction required to do so. Also, instruments with extremely low sounds can "absorb" bandwidth and have an effect on other instruments in higher registers when they play in this low register. In general, you will probably agree that 64-72hz is "bassy enough". Give it a try!



The second thing is an adapative limiter. There are also "mastering" plugs that do a similar thing. This handy thing is like a compressor, but allows you to scale and input against the output, resulting in VERY full sounding, radio-ready tracks. It's quite nice. It also has a nice out ceiling dial which allows you to "master" without going above a certain db. For example, you may want all the music in a game to be at a maximum of -2.5db so that sfx can cut through and be heard. This is a great way to do that.

So what does the final product sound like? Hear it for yourself!

FX discussed in place MP3 audio sample

Few! You can see there is quite a bit to learn and experiment with to produce a final track that sounds much better than what you started with. However, with a few tips from this blog series and other books, videos, magazines, web articles, etc. you should be able to produce something relatively satisfying.

Coming soon...

I'm going to be discussing some of the techniques I use for doing sound design work. Stay tuned!

-John Seguin
Composer/Sound Design/Director
jseguin@seguinsound.com
http://www.seguinsound.com