When Do the Characters Sing? Explain the Rules of a Musical TV Series
When Do the Characters Sing? Explain the Rules of a Musical TV Series
Rowan, alone behind the night desk, sings about how much they hate saying no.
That line is about the fourth page of a pitch. It has a character, a feeling, and a suggestion of music, and it answers almost nothing. Is Rowan singing out loud in the lobby at 2:40 in the morning, with the night porter in earshot? Does the guest waiting at the counter hear it? If the guest hears it, does the plot change? If the guest doesn't, why is it in the show at all?
A reader who has to guess at those questions will usually guess wrong, and then the pitch is carrying two shows: the one you wrote and the one they invented to fill the gap.
So here is the whole job in three questions, applied to every number you propose. What has the scene reached that makes this the moment for singing? Whose reality does the number happen in, and who perceives it? When speech resumes, what is different? Answer those three and you have a convention. Skip them and you have a playlist with dialogue around it.
What follows is a worked example. The series, the hotel, the clerk and both scenes are invented for this explanation, they're prose only, and no song exists for any of it. That last part is a real limit, not a formality: a convention is a dramatic contract you can write down today; music you can't.
The threshold: what has the scene reached when the singing starts?
There's a common shorthand for this — the feeling gets too big for words, so they sing — and it's nearly useless in a pitch, because it locates the trigger in a quantity. Feelings don't have a threshold. Scenes do.
What you're looking for is a blocked move. Something has been asked, promised, or withheld, and the spoken options available to the character right now are all wrong: too small, too late, or a lie. That's the moment of transition, and it's specific enough to point at.
In the invented series, Rowan is six weeks into a night-desk job at a mid-size hotel. At 2:40 a.m. a guest, Mr. Beck, comes down because his airline moved his flight from 7 a.m. to 8 p.m. and he has nowhere to be all day. He asks to keep the room until four.
Rowan already knows the answer. Checkout is noon. The room is sold for tomorrow night to a conference guest arriving at 3 p.m. The hotel's courtesy late checkout is 2 p.m. and no further. Rowan has no authority to conjure a room that has a name on it.
But notice what happens in the eight seconds before the number. Rowan says, "Let me check on that." Then looks at a screen that says the same thing it said before. Then says, "Just one moment." That's the threshold. The two spoken moves left are no and a delay that is a lie, and Rowan keeps choosing the second one, and the scene has run out of room for it.
That's why the number belongs here. Two scenes earlier, nobody had asked Rowan anything, and there's nothing to express. Two scenes later, the answer has been given and the pressure is gone. A number placed at the wrong point in a scene is usually a number whose trigger you haven't identified.
An invitation to perform works as a trigger too, and so does a group's decision to leave ordinary reality together — but notice that those are different entry conditions with different consequences. An invitation implies an audience inside the story. A private, blocked desire implies that nobody in the scene knows the number is happening. You can use either, and you can use both across a series, but each one sets up a different kind of exit.
State who hears what
Sort your numbers into two kinds before you write another page, because they carry different information.
An interior number happens in one character's experience. Other people in the scene don't perceive it, don't hear it, and don't learn anything from it. It can still look and sound enormous — full orchestration, camera moves, other characters appearing inside it — and none of that grants them access to the singer's thoughts.
A shared performance happens in the story's world. It takes up real time, other characters see and hear it, and it changes what they know. Someone can bring it up later. Someone can be embarrassed by it.
The trap is visual. In Rowan's interior number, the imagined versions show Mr. Beck listening, nodding, looking moved. That imagined Beck is a character Rowan is casting, in a conversation Rowan is writing, and he is as invented as the printout of the booking system Rowan produces in the second imagined version. Real Beck is standing at the counter, waiting, and he has not seen any of it. If the pitch doesn't make that distinction legible — in the description, in the imagined dialogue's suspicious smoothness, in the light, in whatever register you choose — a reader will assume Beck was in on it, and every later scene inherits that mistake.
So write the convention down in the pitch in plain prose, once, before the first number arrives:
Rowan's numbers are interior. Nobody in the scene perceives them or learns anything from them.
One sentence buys you the rest of the episode. It also buys the audience a way to watch, which is a real pilot problem: an unfamiliar viewing convention has to be taught, and the cheapest place to teach it is the first number, held slightly longer than it needs to be, with the return to speech unmistakably plain.
A related piece of accounting: an interior number doesn't obey the scene's clock. The real exchange at the counter takes maybe ninety seconds. The imagined sequence can run four minutes of screen time and still return to the same ninety-second conversation. That's a feature, but it means the return is the only thing telling the audience how much real time passed. Anyone standing next to the singer should be exactly where you left them.
Carry something back into the scene
This is where most proposed numbers quietly stop working. The pitch describes the song, the song peaks, and then the next paragraph is about episode three. Nothing moved.
Ask instead: when the number ends, what does the singer do that they weren't doing before?
Rowan's number escalates through four imagined answers. A crisp technical explanation of the booking system. A printout laid on the counter, walked through line by line. A phone call to the manager at 3 a.m., permission granted, a suite opening. And finally the version where Rowan explains their own history — why they took a night job at all — and Beck says the room was never the point. Each imagined version is less about the room than the last.
Then it ends, and Rowan says: "I can't hold it past 2."
That's the whole return. Four words, plus the two things Rowan actually has to give, which are the 2 p.m. courtesy checkout and the luggage room for the afternoon. The imagined versions ran four minutes of screen time; the spoken version takes ninety seconds. The scene is arranged so the decision lands right on the number's heels, which is how the audience reads the number as the place the decision was made rather than a coincidence adjacent to it.
What Mr. Beck leaves the counter knowing: the room isn't available past 2 p.m., his bags can stay, there's a shower room on the mezzanine. What he doesn't know: that Rowan spent ninety seconds deciding, that Rowan is six weeks new, that any imagined conversation took place. The ledger is clean. Nothing has been transferred that wasn't spoken out loud.
In a 2020 episode of the Scriptnotes podcast titled "Sing What You Can't Say," the conversation takes up exactly this stretch of the problem — the move from speaking into singing, what motivates the number in its particular scene, and what happens after it ends. That's one writer's first-person account of one kind of transition, from a craft discussion, not a rule for musicals in general, and it's worth using only for the narrow thing it supports: the entry, the dramatic work, and the exit are scene-writing problems. A number named in a pitch and skipped over is a number nobody has written yet.
Test a second use without moving the goalposts
One number proves nothing. To show a convention has range, write a second scene where the same rule produces a different outcome.
Three nights later, 6:50 a.m., shift handoff in the back office. Two weeks ago Rowan promised to come down at five and cover the desk until the end of Dana's shift. Rowan slept through the alarm and didn't call. Dana stayed until 6:40 checking in a wedding party that arrived on one coach. Dana has never mentioned it.
Same trigger, same reality, same invisibility. Rowan enters an interior number about the apology: an articulate version, a version where Dana says you had a rough week, a version where Dana apologizes first, a version where Rowan confesses more than the apology requires. Then the number ends, and Rowan says: "About Tuesday—" Dana looks up. "—the coffee machine's fixed."
That's the exit. Nothing moved between them. Rowan signs the shift log, writes the coffee machine in the log, and leaves.
Notice what's identical and what isn't. Same convention: interior, unperceived, no transfer of knowledge. Different consequence: the first number equipped Rowan to make a decision; the second substituted for one and exposed an avoidance. A convention isn't a mood. If every number in your series has to do the same emotional job, you have a formula, not a convention — and readers can feel the difference.
Now the version that breaks. Suppose the scene ends with Dana saying, "It's okay. I forgive you. I know you didn't mean it." Under an interior convention, Dana has heard nothing. Dana cannot forgive words that were never spoken in the world of the story, and the pitch has now given a character knowledge the convention says they can't have.
There are two honest repairs, and they cost different things. Keep the convention and change the consequence: Dana never learns, the gap stays open, and the scene's job is to show the audience something about Rowan rather than to resolve anything. Or change the convention deliberately: make the apology a shared performance, sung in the room, in real time — and then accept the bill, which is that Dana knows, anyone else in earshot knows, and the relationship has to move in the next scene because of it.
What you can't do is let Dana hear it because the story wants them reconciled by episode four. An exception that rescues a plot is the fastest way to make a convention unreadable, because the audience can no longer use the rule to predict anything.
The same logic governs mixing conventions on purpose. If this series also wanted a performed number — a pianist in the hotel bar on Fridays, where the singing belongs to the room and to everyone in it — that's available, and it's a legitimate second register. It also raises the price: the pilot now has to teach the audience to tell the two apart, and once an in-world source for music exists, viewers will look for one in numbers that don't have it. Mixed conventions work when their differences are marked early and never blurred for convenience.
The paragraph that goes in the pitch
Everything above collapses into one short statement you can paste into the document, followed by the compressed scene descriptions that demonstrate it. For the invented series:
Numbers in Night Desk are interior. They happen inside one character's experience, are not perceived by anyone else in the scene, and never change what another character knows. A character enters a number when the spoken move available in the scene is smaller than what the scene has accumulated — an ask they can't grant, an apology they can't start. The number plays out in imagined space and can run longer than the real scene. It ends with a return to speech, where the singer makes a choice or fails to make one. The consequence stays inside the singer.
Then two short descriptions, entry to exit:
Scene one. Entry: Beck asks to keep the room until four; Rowan has checked a screen that already answered, twice. Number: four imagined answers, each more elaborate and less about the room. Exit: "I can't hold it past 2," plus the luggage room for the afternoon. Beck goes upstairs knowing the no, the luggage room, and the shower room on the mezzanine — and nothing about the number.
Scene two. Entry: 6:50 a.m. handoff, an apology owed for two weeks. Number: an imagined apology that keeps improving. Exit: "About Tuesday—the coffee machine's fixed." Nothing changes between them; the audience learns something about Rowan instead.
Read those back and check whether a stranger could tell, from the convention paragraph alone, whether the person standing next to the singer knows anything afterward. If they can't, the paragraph isn't finished.
Two boundaries worth keeping visible. First, this is an invented example in a draft, and it hasn't been tested on anyone; whether an audience follows it is a question you answer by writing and showing the scene, not by argument. Second, a musical series in a pitch usually arrives with reference tracks. A reference track tells people something about sound. It isn't the rule of reality the numbers obey, and it isn't an asset — who writes the songs, what they cost, and what you're allowed to use are separate, unresolved problems that a convention paragraph doesn't touch.
The rule is the thing you can settle now. Singing starts when speech has run out of honest moves. State whose head it happens in. Bring one changed thing back with you when it stops. Do that twice, with different results, and your pitch has a world instead of a wish list.
Frequently asked questions
What three questions does the article say every proposed number should answer?
What the scene has reached that makes this the moment for singing; whose reality the number happens in and who perceives it; and what is different when speech resumes. If those are unclear, the pitch reads as a playlist with dialogue around it.
What makes a scene ready for a song, according to the blocked-move idea?
A blocked move: something has been asked, promised, or withheld, and the spoken options are all wrong—too small, too late, or a lie. In the invented Night Desk example, Rowan cannot grant the late checkout and keeps choosing delay-that-is-a-lie until the scene runs out of room.
How is an interior number different from a shared performance?
An interior number happens in one character's experience; others do not perceive it or learn from it. A shared performance takes up real time in the story world, is seen and heard by other characters, and changes what they know. The visual trap is making an imagined listener look as if they were actually present.
What must happen when the number ends?
The singer should do or fail to do something they were not doing before. In Rowan's first number, the exit is 'I can't hold it past 2,' plus the courtesy checkout and luggage room; in the second, the apology becomes an avoidance—'the coffee machine's fixed'—so nothing moves between them.
What common mistake breaks an interior convention?
Letting another character act on words they never heard, such as Dana forgiving an apology that existed only inside Rowan's number. The honest repairs are to keep the convention and change the consequence, or deliberately make the number a shared performance and accept that other characters now know.