ADR

ADR (Automated Dialogue Replacement). . .
requires three basic steps for creating newly recorded dialogue to sync with picture:
Set up the dialogue loops . . . Record the talent & effects . . . Mixdown the audio program

The process of replacing dialogue in a film or video is a craft unto itself. We have all seen it done poorly, even in national ads and movies released for international distribution. This page provides an overview of our process for doing it well, and it will also make a good primer for recording engineers wanting to develop a suitable technique of their own. Two video samples are included. The one at the top of this page is edited, raw camera footage with sound from the camera's microphone. The sample at the bottom of this page demonstrates the same video clip after re-recording of the dialogue in a studio, mixdown with ambient effects and sync with the picture. The second sample also demonstrates how the picture quality can be polished a little using the color corrector plug-in with Final Cut Pro.

It is true that the basic steps for ADR are fairly simple for any experienced recording engineer. But film-sound engineers pick up techniques on their own with practice that can make ADR sessions run more efficiently. And as with recording music, a high level of focus is required by the performer. So anything an engineer can do to make things more comfortable for the talent will help shave a lot of time off of the recording session.

Setting Up Loops

If you are using a professional quality digital audio software such as Digidesign's Pro Tools or Apple's Logic Pro, then you are equipped with a perfect tool for ADR. I have been using Logic since, well. . . since before the application was even called Logic. It was exciting for me, as I'm sure it was for all recording engineers working steady through the mid 1990's, to witness the birth of personal computers with enough processing power to record audio at a professional level. While in college, my studies in sound and music production for film preceded this technological revolution. So I was required to learn sound editing for film using techniques which dated back to the 1920's. But don't let the sophistication of modern computers fool you; those old techniques still apply!

For example in setting up loops that your talent will hear during the recording session it is useful to think of a loop of film, a length of magnetic film ("mag" for short) containing dialogue in this case. The loop is a piece of this film that has either been re-recorded on a magnetic recorder from the original audio recordings or a piece of mag cut out of a longer piece, a longer piece that contains several lines of dialogue spoken by the actor in the scene. So the sound editor's first job in the established technique was to take instruction from one of the movie's producers concerning which lines needed to be replaced in the scene, make loops of those lines by splicing the start of a length of mag to the end of itself, and get the set of loops ready for the ADR engineer to place on an audio playback machine (mag player) that would sync with a projector to display the picture of the scene to the actor (or voice-over talent) during the ADR session. Whew, sounds complicated! It was; although studios that specialized in this type of work previous to the advent of computer audio became able to cover a tremendous amount of work in a day. If you never had an opportunity to do this type of session in the pre Pro Tools days you have no idea how lucky you are! Or maybe you do.

In Logic Pro, as with most other modern multi-track recording software, it is simple to find a start and end point in your recording and set it to loop seamlessly. Logic has a toolbar that includes a switch to turn looping on with a single mouse click. In the transport control window there are sections that allow quick designation of two song positions. Note that the line, or cursor in the arrange window of this type of software is commonly referred to as a Song Position Pointer, since these applications were originally designed for music production. So find the beginning of a line of dialogue that needs replacing and type that SPP (Song Position Pointer) into the "Start" or "In" box of the transport window. Find the end of this line where you would like recording to loop back to the line's start and type that SPP into the "End" or "Out" box. Click the "Loop" feature in the appropriate toolbar. Voila, you are ready to loop the line. With good recording levels set on your mic preamp and interface, start the loop which should be fed to the talent's headphones at a level comfortable for them. They need to be able to hear themselves in their headphones just well enough to know when they are in sync with the loop. As the rhythm of their speech locks in with the rhythm of the loop, click the record button in the transport window or your appropriate keyboard command to start recording. What Logic Pro and many other applications do at this command is to begin recording the loop and, at the end of each automated pass, record the next performance of the loop to a fresh track in the arrange window. Once you, the talent or the producer decide that a few good takes have been recorded during this automated process, stop recording and set up to loop and record the next necessary line of dialogue. Once the session of recording the lines is done and the producer and talent are satisfied with the takes, the talent and producer are free to leave and schmooze. That just leaves you, the dedicated engineer, to follow up with checking the takes, editing out noise, adjusting E.Q., balancing levels, fine tuning position of each line with the video, finding the perfect reverb, mixing the audio with Foley, room tone and other effects . . . well, you get the idea. I work occasionally as an actor, so I understand these processes from both sides of the tracks.

In some cases there are many lines that need replacing. It may even be necessary for every line in a scene or the entire movie. Such was the case in the video samples on this page . . . more on that later. In this type of situation it is only fair to the talent that you set up all of the loops in advance, even if you don't expect to get all of the ADR done in a single session. In Logic's transport window there is a button to open the Markers window. And the use of Markers is how you can set up a series of loops before the session. Different applications such as Pro Tools, Digital Performer, Vegas and others will have different keyboard commands and menu items for creating Markers. I am referring to Logic Pro in this explanation since it is the application I use most often. When you find just the right start point for the first line of dialogue you plan to re-record, set the SPP there and simultaneously key Control-Command-up arrow (Logic only runs in the Mac OS, by the way). Logic sets the Marker in the Time Line as "Marker 1", then continues counting them up sequentially as you add more. It also rounds off the position of the marker to the nearest beat, so if your Marker point needs to be placed somewhere between beats, you will need to use the menu command "Create without rounding" instead. You can also set up a key command for "Create without rounding", if you like, because Logic's "Key Commands" preference (under "Logic Pro>Preferences") allows you to customize your key commands just about any way that suits you. Now when you set your end-of-the-dialogue-line Marker it will be "Marker 2". Continue with this process for every line of dialogue that needs to be looped. When you are finished, all of the Markers will appear in the Time Line as well as a list in the Markers Window. The odd numbered ones are start-of-loop points, and the even numbered ones are end-of-loop points. Now enable Logic's "Loop" feature with a mouse click. Congratulations. If you had an especially large amount of loops to set up, then you probably just created them in about one-tenth the time it would have taken a sound editor back in the day. It's a good idea to use some of that time you saved, however, to review all the loops. If they start and loop in a rhythm that feels comfortable for you to speak along with, then they will probably be fine for the talent as well. But be sure to let the talent know before the start of the session that you can easily modify the start and end of any dialogue loop that does not feel comfortable to them (another perk not available back in the day).

Video Monitoring

Although video monitoring for the talent is not an absolute necessity with the looping system of ADR, it is a major help. This becomes especially evident when replacing lines of dialogue where the actor's face is visible in the scene. The human face is capable of a great variety of expressions, even when you are seeing nothing but lip movements. These are very helpful cues to the talent as to exactly how a line should be performed for the best match and best expression. Even subtle movements of the actor's head and body can be useful cues to the voice-over talent. Here again modern applications like Logic Pro help out. These types of applications allow you to watch the movie Picture-In-Picture style while using the recording features. The movie will run in sync with Logic in either Record or Playback mode. And there are very fine adjustments for "Movie Start" which will allow you to slip sync between the movie and the audio playing in Logic to achieve perfect sync. Other features include the ability to size the video window, transport controls on the video window that can lock with the SPP, and a "Contiguous Sync" feature which makes Logic control the playback speed of the movie to keep newly recorded sound and the picture in sync at all times. When I use Logic to create dialogue loops, I usually set the "Video Output" (under "Video" preferences) to "Window" so that I can simply watch the movie on my computer monitor. But when I set up to record the voice-over talent, I change this setting to "Firewire". I have a Firewire-to-Composite Video unit made by Canopus that will then allow me to send the video signal to any monitor or TV set with a composite video input. Or the signal can be run to an A-V receiver for distribution to several monitors at the same time. It is very helpful to set up a separate monitor for the producer or director so they do not have to look over anyone's shoulder during the ADR session.

Ambient Effects (a.k.a. - Foley)

Ambient and incidental sound effects in a scene are sometimes the most fun to record, and they do not have to be very complex to work well. In fact the more subtle they are the better. Simple noises like an actor smacking their lips between lines, rustling of clothing, a puff of a cigarette, etc., benefit the scene by making the replacement dialogue look and sound more like it was actually recorded with the action. If you own the microphone you are using for the ADR session you are better off, because it is good to use the same mic for subtle sound effects that you use for recording the talent's voice. If you are instead renting an expensive mic for the session, be sure to plan on taking care of as many of these sound effects as possible while you still have the mic. I find that voice talents are very good at these effects as well, and they usually don't mind spending a little extra time in the studio to record them. I have also found that, even in situations where the location sound was recorded poorly, brief ambient effects can be cleaned up a bit in Logic and used in the sound effects track for the final sound mix. An example of that in our Persona Redux project was the sound of Tzvetana flicking on a cigarette lighter (which does not appear in the included sample). Another finishing touch is to record some Room Tone. This is done by simply recording pickup from a high-quality microphone in a room that is equal in size to the room which appears in the scene. Record about one full minute of silence in the room, place it on a track of its own, and set it to loop continuously throughout the scene. Logic's loop button will enable specified audio objects in the arrange window to do this independently of other objects (independently of the Markers Looping feature we described earlier as well). When working on the final sound mix, you will find that it takes just a little of this Room Tone mixed in to help marry the replacement dialogue to the picture.

Sound Mixing

If you have performed a good deal of mixing for music productions you will have a head start on being able to perform good mixes for a movie soundtrack. You will have spent much time doing critical listening, usually without the benefit of being able to watch anything other than the actions of your software on a computer monitor. The addition of live action or animation to watch while mixing may seem to complicate things, but in most cases it helps with your decision making. I would estimate that by the time a final, satisfactory mix is achieved on one song consisting of tracks by a group as simple as a five-piece band, that the mix engineer has made anywhere from 1,000 to 10,000 important decisions (and perhaps thousands of other minor ones). So this stuff takes patience and focus. Continuing practice helps it get better and easier, like playing a musical instrument.

In a movie sound mix, you are basically trying to get all of the separate sound elements to sound like they have a reason to be there. I point this out first, because you may very well end up with some components of sound that are getting in the way of a sensible and effective soundtrack. Music for scenes is a special art itself, so it is important for the composer and director to attend the session if it is a final sound mix that will include the music cues. If you are working alone on just the dialogue mix, then spend some time listening without watching. When the sound of dialogue combined with sound effects starts to tell the story to you when your eyes are closed, then you know the mix is close or already there. Now look at the picture again while listening. Make more adjustments. . . . With patience and concentrated effort you will get the perfect mix for the scene. If you are working on a full length feature . . . cool . . . just twenty-nine (or thirty-something) scenes left to mix!

When all of the dialogue for scenes is mixed down to satisfaction, and the same has occurred for the sound effects track and music tracks, you will have some options on how to proceed with final sound mix. This stage of the soundtrack creation will usually consist of the three major components: DIALOGUE - SOUND EFFECTS - MUSIC. These three components are usually stereo mixes that are referred to as Audio Stems. If you are planning to plug the three Stems into the audio tracks of an application like Final Cut Pro, then achieving just the right balance between these Stems can be effectively done there or in most other NLE's. This also gives you the opportunity to watch the movie at full resolution while this critical step of balancing is done, and seeing the picture in full glory will influence mix decisions. In a major post-production facility, such as Todd-AO studios in Los Angeles, the picture will be projected onto a large screen equal in size to most neighborhood cinemas. But most low to mid-budget features and many animated features (especially for TV) are mixed in smaller studios where the largest screen available is, perhaps, a 40" LCD monitor. If you are fortunate enough to have a manager friend at your local cinema allow you to view your project on the digital projection system there, it helps. We have been able to screen our Persona Redux project this way, and it has benefitted our decision making. I actually like to do as much of a final mix in Logic Pro as possible. Logic includes the ability to output stems for a 5.1 or 7.1 surround mix (with various options). Surround sound is practically another world for mix engineers, but it is built upon all of the best practices of single track and stereo record engineering. The surround stems also need to be properly encoded before a movie file can be either projected in a cinema or written to a video DVD. Newer versions of the Toast disc writing software can help with this.

A Few Notes About the Video Samples. . .

In collaboration with film maker Thomason Pollard, cinematographer David Simmons, and actors Tzvetana Denkova and Nancy Mills, I have assisted in producing a study of Ingmar Bergman's classic film, Persona. Thomason wanted to better understand the process that Bergman used for creating his masterpiece, and did so by re-filming and re-recording a pair of scenes from it, line for line, shot for shot. We referenced a good English translation of the script and, during the course of a full day of work, shot everything needed to create an eleven minute movie. My audio pickup during production sounded good on headphones, but I discovered after the shoot was over that the camera's (Panasonic DVX-100B) microphones had recorded the audio instead of mine! It was not usable sound, as you can easily tell from the raw sample at the top. But instead of throwing in the towel, Thomason and I realized it would be a good opportunity to re-record the dialogue using ADR if Tzvetana was willing to do it. Nancy's character, Elizabet Vogler, never speaks, and those of you familiar with the movie will know why. Tzvetana graciously agreed to do the ADR, and with just a few brief sessions we re-recorded all 85 of her lines. She did a wonderful job with what was, indeed, a tedious process. I used an Oktava MK-219 microphone (one with modification by Michael Joly) to record the dialogue during ADR and used a process similar to what I outlined earlier in this article. Thomason and Tzvetana performed the sound effects while watching a video monitor to help get a good match. A track for Room Tone was mixed in and a stereo reverb with a ring time of just under 0.5 second was used on the dialogue track. A touch of color correction was used on the picture in Final Cut, mainly to brighten highlights, midtones, and bring out some of the shadow detail. The demo is compressed using Apple's H.264 codec.