Projection

Timing titles to a live conductor rather than to timecode

The score is the timecode, the performance moves every night, and a title that arrives early has told the joke.

For a separate operational view of calls, time, ownership and crew activity, see this Monitask guide.

A film subtitle is locked to a timecode and is identical at every screening. A title in a live performance is released by a person watching a score, and the performance underneath it changes every night. That difference is the entire discipline.

The score is the timecode

Preparation means marking a copy of the score with an in-point and an out-point for every title, so that the operator is following music rather than counting seconds. The in-point is usually a specific beat or word. The out-point gets less attention and causes as many problems: a title left up after the thought has moved on gets read a second time, and the audience is reading old text while new music happens.

Instrumental passages are where discipline shows. Nothing needs to be displayed during them, and clearing the panel returns the audience's eyes to the stage for the length of the ritornello. A panel that never goes blank trains people to keep looking at it.

Early is worse than late

A title that arrives a beat behind is a minor irritation. A title that arrives ahead of the line it translates has told the audience what is about to be sung. In comic repertoire this is fatal: the laugh happens during the setup, the performer delivers the punchline into a room that has already had it, and the timing the company rehearsed for weeks is gone.

Where a cue is uncertain, the operator should be late rather than early, and that instruction belongs in the title script rather than in the operator's judgement at the moment.

What moves underneath

Tempo is the obvious variable and the least troublesome, because the operator is following it anyway. The real hazards are discontinuities: a repeat taken when it was cut in rehearsal, a verse dropped, a page turned early by a nervous singer, a cover performer working from a different edition.

The requirement that follows is navigational rather than technical. The operator must be able to find their place in a hurry, which means a script whose cue numbers are visible against recognisable musical events rather than against bar counts alone, and a display system that can jump to an arbitrary title rather than only step forward.

The sung phrasethe point of the lineCued with itin with the first word, out before the next thoughtCued earlythe audience has read the joke before it is sung
Figure 1Cued with the line, a title supports it. Cued early, it delivers the point of the phrase before the performer reaches it.

Language decides how much time there is

The same phrase takes different numbers of characters in different languages, and the music does not adjust to accommodate that. A dense inflected source rendered into a language that needs more words leaves the operator with text that cannot be read in the interval available, and the resolution is editorial rather than technical: cut the title, not the display time.

This is why title scripts are written by someone who knows both the language and the repertoire, and why a translation commissioned as prose and then chopped into panels reads badly. The unit of composition is the panel, and it has to make sense alone, arrive whole and leave before the next thought.

Automatic score following, and what it does not solve

Software that listens to a performance and tracks its position in a score exists and is used, and in stable repertoire with a clean audio feed it works. It handles tempo variation, which is the part a human also handles.

What it handles less well is the part that actually goes wrong. A cut that was not in the reference, a passage the orchestra takes at half the expected speed, an interruption, applause in an unexpected place. Systems recover from these to varying degrees, and the recovery is exactly what cannot be demonstrated in a controlled test.

The honest position is that automation reduces workload and does not remove the requirement for someone who can take over immediately. That person needs the score, the script and a manual control, not a screen showing what the software thinks is happening.

The operator is in the calling structure

Titles are cued from a position with sightlines to the stage, headphones on the show channel, and a place in the stage manager's calling script. Where they sit outside that structure, the predictable failures follow: the panel stays lit through a scene change, nothing clears at the interval, and a hold on stage leaves a title frozen over an empty set.

The role also needs rehearsal on the stage with the orchestra, not only a run against a piano. Cue points that were unambiguous in a rehearsal room can be inaudible under a full orchestration, and the operator discovers which ones only by sitting in the position and running it.

What we cannot verify

Accuracy and recovery claims for score following software come from the companies and research groups that develop it, are measured against particular repertoire and recordings, and are not comparable between systems or predictive of a specific production. We reproduce none of them. Practices for preparing title scripts vary widely by house and by language pair, and what is described here is common practice rather than a standard; no standards body governs this work.

The short version

  1. The score is the timecode, and every title needs an out-point as well as an in-point.
  2. Clearing the panel during instrumental passages returns the audience to the stage.
  3. A title that arrives early has delivered the punchline before the performer.
  4. Where a cue is uncertain, late beats early, and that belongs in the script.
  5. Cuts and repeats, not tempo, are what strand an operator.
  6. Automation reduces workload; it does not remove the person who takes over.

Further context

For a primary, standards or institutional reference, see the ITU Radiocommunication recommendations.