Sound effects in musicals are an odd beast.
Well: just as in straight plays, there are plays with many sound effects, and plays with few. In the latter category come those drawing room comedies where all the action takes place inside a box set and nothing happens but an occasional phone ring. These are the designs you can (if you'll forgive the pun) phone in.
There are also plays, such as any of Shakespeare's, in which the play could be done with strong sound effects, but generally aren't -- mostly because most directors don't think in terms of sound. The same play, then, can be reduced to "A bear roars off-stage," or in other directorial hands be a through-composed seamless and continuous panoply of quadraphonic ear candy.
And then there are plays in which the environment or elements in the environment cry out so for sound effects it is hard to find a director who won't want them.
Musicals embrace all three of these worlds. But then they add another dimension; that they are, well, musicals. There is singing and a band playing for much of the running time.
This hits your sound designs with three constraints; intelligibility impact, limited space, and necessary timing.
Take the last first. In "The Jungle Book," Shere Khan needs to roar several times. All but one of those roars are in the score. Which is to say; they have to happen at a specific moment in the orchestration, and occur on the beat. In "Into the Woods" there is a baby cry that is scored in the Entre-Act. In "Seussical" the opening of the Egg has to occur over and within four specific measures of the score; heard before the first bar and concluding with the last.
There are two techniques that help greatly with this. The first is to record the rehearsal. I used to do it on mini-disc, now I just use my laptop. Sitzprobe is a great time to do this. This is also a great technique for certain cues in straight plays as well. Record the actual scene as sung/performed by your cast and/or orchestra.
Back in the early days I'd use a stopwatch and jot down time through the scene in question:
0:0 thomas in
0:5 "what'cher doing?"
0:12 falls in water
1:33 approx scene end
Now I take the recording from rehearsal and make it the top track in CuBase. I can then do a rough synchronization with that as a guide track.
When I did this for the dogfight for "Your a Good Man, Charlie Brown" I knew that I was not going to depend on the scene staying in lock step to the exact time they'd spent in rehearsal. I designed a dozen cues in total that would come in, or cross-fade, at specific moments in the action.
So this is how I built that show; I started by making a rough cut of the entire scene as it would be heard in performance. Then I saved the original file and began to edit; for each sub-cue, I muted the elements that would happen in the next cue, and extended the events that would happen in the current cue, and rendered a sound file that was long enough to cover any change in the timing. Then, in Qlab, I built a show that would fade out or cross-fade the previous cue as each new cue was played. This gave me the ability to do a constant engine sound of the Sopwith Camel that could still react to the power climb and stall out that happen at various moments in the music. The ack-ack, machine guns, and so forth were layered on top to be played simultaneously with the engine noise bed.
The same technique is wonderful for composing musical underscore for a scene in a straight play.
But back to musicals. Another trick I've been using more and more is to hand over control of a sound effect to the band. Since of course they know where they are in the music at all times, and they can see the conductor's baton to get accurate tempo.
In many cases, it is most appropriate to leave behind your own ego, and hand the sound over literally; let the percussion player do something instead of making it a recorded sound effect.
If it has to be recorded, like the baby from "Into the Woods," I've been handing a MIDI device to conductor, keyboard player, or percussionist. A small sampler (like the Boss "Doctor Sample") does the trick. A laptop and the right software would too. Or stick a keyboard or MIDI drum pad down there and route it to a sampler, soft sampler running on a laptop, or the show playback software.
Or you can have a button. I created my own MIDI button a while ago using the Arduino as a platform. Since it is detecting a state change in 5v TTL, I can run it through two hundred feet of XLR cable without worry about dropping signal. The conductor or drummer gets a box with a big red button on it. The wire leads to the brain box in the booth, and Qlab gets the signal via MIDI to trigger the sound effect.
Ineligibility is shorthand for the problem that in a musical, there is already a lot of sound going on. Singing, the orchestra, the sound of the actors in motion, moving scenery -- all of that conspires to make a congested sound environment. Often as not you will simply have to omit a sound effect because it would get in the way of the audience being able to hear the more important things (like the lyrics.)
This runs into the third constraint I mentioned, in that it also compels you towards sound effects that are short in duration, simple in form, and that have a limited frequency range.
That last constraint is that, again, with all the other noise going on nuances are going to be lost. And with the fast-moving nature of a musical, and the necessity that the effect be finished and over before the next verse of the song comes in, the pressure is towards sounds that are instantly identifiable and that can speak their piece and get out of the way.
The musical is rarely a place for nuanced, expressive sounds, or complex environmental sounds, or detailed story-telling. What you want instead are sketches, cartoons; sounds pared down to the minimum that expresses the idea.
As a for-instance, I have a "Palm Beach" cue in my current production. Surf and seagulls. The cue was constructed so the seagulls make a single loud establishing chatter, and there is one loud wave, then the birds fall silent and the waves drop down to a lower level. So the sound only plays at volume for 3-4 seconds before dropping down to an inconspicuous background. The surf, too; I started with the most generic surf sound I have, and equalized it to bring out as much wave froth as possible and take out pretty much everything else.
Played baldly, the cue sounds horrible. Seated into a scene that starts with loud orchestral underscore and seques immediately into a song, it works quite well.
My first stop for these cues is to find the most generic lion roar or train whistle I can. The art is to find that sound that even played at half-volume over a phone will be instantly identifiable. Hollywood is very good at this. The thing is, those "instantly recognizable" sounds are rarely anything like what those things actually sound like. No dog in the world ever said "bark" after all (or "wan," for that matter.) Eagles do not sound like a red-tail hawk. Most monkeys do not make the stock "Eee, eee, eee!" sound. Snakes inhale as well as exhale (aka hiss). And of course a real gunshot sounds like a stick slapped on a hard surface and a real punch lacks, well, punch.
The next step, besides the inevitable equalization, is to trim the cue for time. Often I've had to use Time Bandit or similar to shorten the cue. There has never been a musical with enough time for an airplane to actually make a full pass; instead I take the best sound I have, cut a section out of it, and apply pitch shifting, panning, and volume envelope to create a new and much shorter fly-by sound.
It is pretty much part of the above, but many of these ultra-short cues will still be compound cues. It has to do, again, with audience expectations. Four seconds of "chuff-chuff-chuff-chuff" might not be enough for the audience to run through all possibilities in their minds and arrive at "Oh, a train." But if you add a (very shortened!) train whistle on to it, you can have a four-second cue that they will recognize is supposed to be a train. It is, as I've been saying, a sketch of a train, a cartoon, the kind of outlining a landscape painter does to create the illusion of a fir tree with a dot of paint and a brief brush stroke. Given time to think about it, the audience would find that a fairly ridiculous excuse for a train sound. But in the context of a scene change in a musical and all the artificiality that entails, it works.
Games are a good place to see this sort of art of compression. Not modern FPS games -- RTS and TBS games like the Civ series, the earlier Warcraft, Sim City, and so forth. Think of those games where you click a "build farm" icon and get a two-second sound burst that says "Hi, I'm a farm."
So here's the rough process for designing these things:
1) Read the script and talk to the director (the first design/production meeting, an interview, an email -- whatever it takes to get a basic idea of where they are going with the show.) This is not a time for minutiae; this is a time to learn if the show is realistic or not, what period it is, how much the director is open to sound effects, whether it is the kind of show where live practical effects or orchestral sounds are more appropriate, and so on.
On a recent production of "Your a Good Man, Charlie Brown" the idea was presented in the first meeting that the entire show occurred as if on a playground, and that everything that happened was being acted out by children who were playing the environments as well as the parts. This led into discussion of "detoxifying" the "Red Baron" scene (trying not to have a lot of war sounds in a play with and for children) -- and a moment of inspiration when we realized the direction to go there was to play the entire scene as if children were playing war on the playground and to make all the airplane and battle noises from actual recordings of children's voices.
2) Spot cues. I like to take the script to my favorite cafe and have a leisurely brunch while marking it up. I usually make a complete pass of the script sticking colored tape markers where ever there is something that might be a sound issue. I even use several colors of tape; one for scene changes that might involve a background or environmental sound, one for spot cues, and one for potential vocal processing -- such as the Ghost of Christmas Past, or The Knight of the Mirrors, or perhaps a character who is temporarily down a well.
What you are looking for are several things. You are looking for artifacts on stage that might not make the right sound by themselves...a prop gun, for instance, that might need to have an effect when it is fired. Or a vacuum cleaner, which might not be practical and would again need a recorded cue.
You are looking for changes of scene or time that could benefit from something to underscore them. For instance, one scene may be "that evening." To emphasize that the time of day has just changed, throw in some crickets. You are also looking for places to enhance a setting. The sewer in Guys and Dolls is a giant set-piece scene that wants to be amazing. Add some drips and steam to make it come to life.
You are looking for things that aren't on stage but are either implicit or explicit parts of the action. In the latter, "Listen to the howling of the wolves!" says a character. You probably want some wolves for them to listen to! In the former, the scene might take place on a street corner, and even though no-one mentions the cars passing by, to leave them out of the sound picture would be odd.
And, of course, you are looking for the explicit sound effects described in the script. Please be aware that in some musicals, there are effects that don't get mentioned in the script but that DO appear in the written score.
3) Tentative cue list. This is a list that shows page number, kind of cue, and a brief description. I use my tentative cue list as a reference for coordination with the other departments; "These church bells heard on page 33. Are those a sound effect, or is that something the orchestra is playing there?" Or, "This door slam on page 15. Are we going to have a practical door for that or do you want a sound effect?"
So my list is not just sound effects, but also includes even those things like starter pistols, crash boxes, and actors yelling lines from the wings that I am pretty sure are not going to be pre-recorded. Because you want to ask. And if it turns out it is a sound effect, you've already thought a little about how to do it.
This list gives the director a firmer sense of how much sound you intend to use, how realistic you are going with it, and so forth. And it gives them a place where they, too, can say "I've added a bit in rehearsal and we need a siren sound on page 3 for that," or "We've cut the business on page 114 and there is no sewing machine anymore."
4) Pulls. I make up a pull list. This is not the list of actual sound CUES. This is a list of the individual sound EFFECTS. For instance, effect might be "Ocean liner hits iceberg." I know I have plenty of stock water and boat sounds, and those will be easy to pull, but the pull list reminds me I need to search for something that says "Iceberg." As much as I can, I break down the problem and come up with tentative solutions so instead of doing giant library searches for "Titanic disaster" I can do a specific search for "ice sliding across metal" instead.
My method has pretty much always been to start a directory for the show and make sub-folders for major cue groups (On "The Jungle Book" my sub-folders are "Monkeys, wolves, tigers, elephants, jungle backgrounds, water, sand and weather, other.") Then I copy from my SFX libraries so all the files are right there, consolidated.
I need to look for gaps and go purchase/record/synthesize new sounds. And I need to audition sounds -- playing one tiger roar after another to narrow down the best roars for that particular production. And the pull list is also a track of work completed versus to be done.
5) Trials and tailoring. This is where you want to be in the theater if possible, and in the rehearsal hall if time permits. You want to listen to the sounds on the actual speakers and as much as possible (given the schedule) within the show context. The biggest problem here is that orchestras do not get integrated until very late in the rehearsal process. You have as little as two days with the full orchestra until you go in front of an audience.
Trouble is, you won't really know if the sound is going to work until you hear it in place. Worse yet, the director has no idea. Some directors will try to second-guess you -- and with a musical to manage, they don't have time for discussion. They are liable to say "It isn't working, cut it" without giving it a chance to be heard in context. These are the times that professionalism, trust, and a soft touch are most necessary.
To get through a musical you have to work fast and stay fluid. It may work out best to have a bunch of sounds in progress and to play from the actual work files during rehearsal instead of from fully rendered cues. Because then you can make changes on the fly and even try it again if the rehearsal permits. Be prepared in the fast pace of tech for a musical to have days of work turn into a discarded file because the scene works better without the effect -- or to have last-minute requests for something new and ridiculous. (My favorite at the moment is the last-minute request for a sound of "..the snakes getting zapped by love.")
6) Installation. This is where you shift sounds from "stuff you are playing with during rehearsal" to actual numbered cues entered into the Stage Manager's book, called to an operator, and executed from show software.
I've done sound effects both ways. I like the control you have as designer to chose where exactly to hit a sound effect for best integration with the music and flow of the show, and pinpoint volume control to seat it into the mix, but running sound effects whilst mixing microphones can get rather exciting.
The best solutions I've had are using a MIDI keyboard or my MIDI button so I can hit the sound effect without having to move my hands too far from the faders I am mixing. Still, there are far too many times when you've got a complex scene with multiple actors getting into each other's microphones, a bunch of fast entrances, and a bad microphone you are trying to track down before it blasts the audience with noise, and the sound cues end up getting skipped in the fuss.
My second preference would be to have an operator running effects off their own copy of the script. They can then nuance their performance to the flow of the show, and they can be given "hot" buttons to fire off mickey-moused effects that would otherwise be too fast or too risky to design in implicitly. For instance, ad-lib train whistles, gunshots, thunder crashes.
However, basic theater politics holds that you will only get a volunteer or intern to do this, and they WILL be subservient to the Stage Manager, so it will be a very hard fight to get them off headset and given the freedom to take their own cues. As much as possible, theater will attempt to nail down any chance at spontaneity and will turn your sounds into discrete and distinct moments that will be fired at pre-determined moments linked only to the words in the libretto -- and not to any real sense of what is happening on stage.
Fortunately, a musical is generally tight enough in timing that this works for most of the cues. A good Stage Manager will slide the timing to make up for late entrances or a change in orchestration the pit just came up with or other changes in the flow of the moment that make the sound cue not work as well in the original spot.
What you have to do is (and this happens for sound design in straight plays as well) is to break into that mindset that says the words are everything, and that "Actor says the key word" equals "sound effect happens here." It is also true on lights -- sometimes it is a surprisingly uphill battle to get it understood that where the cue was designed to happen is as the actor crosses the threshold -- and the fact that they are saying their line a beat earlier then they had during rehearsal has nothing at all to do with where the CUE belongs.