A role-playing game with a fully voiced script is one of the largest recording projects in entertainment, and it is produced under contract structures that shape the writing itself.
Sessions are short and scheduled in blocks
Voice performers are typically booked in sessions of a few hours, and a single performer's entire contribution may be spread across several such blocks over months.
Because studio time and performer availability are both scarce, lines are recorded grouped by emotional register and by scene type rather than in story order.
Performers consequently deliver a scene without having played the sequence around it, which is why context notes from the director matter so much to the result.
Union and non-union terms differ substantially
American voice work may be produced under a union agreement or outside one, and the choice affects session length limits, vocal stress protections and residual arrangements.
Union agreements set minimums and working conditions, particularly around shouting and vocally damaging performance, which large role-playing games require in volume.
Smaller American studios often work non-union for cost and scheduling flexibility, which broadens the casting pool while giving performers fewer standardized protections.
Retakes are a contractual question
Games change during production, and a rewritten scene means bringing a performer back for lines that were already delivered and paid for.
Whether that is a new session, a pickup at a reduced rate or covered by the original agreement is decided in the contract rather than in the moment.
Teams that expect heavy iteration therefore structure agreements to allow pickups, which is one reason narrative scope is locked earlier than other systems.
Casting decisions bind future projects
A recognizable voice becomes part of a character's identity, and recasting for a sequel is noticed immediately by an audience that has spent dozens of hours listening.
Studios manage this with options and availability terms, but a performer's schedule years later is not something either side can guarantee.
The risk is real enough that some teams deliberately limit how much a character's identity depends on one performance, particularly for player-facing protagonists.
Synthesis has made terms the central issue
Recorded performance can now be used to train systems that generate additional speech, which raises the question of what a session's recording is licensed for.
American performers and their representatives have pushed for consent and compensation terms covering that reuse, and those clauses now appear routinely in negotiations.
The technical capability is settled. What remains contested is the contractual scope, and that is being written project by project rather than resolved at once.