An English dub is recorded against animation that is already finished, which reverses the usual order of production. That single fact governs most of how casting works.
The performance is fitted to existing mouth movement
Original Japanese dialogue is generally recorded to picture as well, so the animation carries mouth flaps timed to lines that already exist.
An English script must convey the same meaning in roughly the same number of syllables, landing at the same moments. Adapters rewrite lines specifically for that fit.
Casting therefore favors performers who can hit precise timing repeatedly, since a beautiful reading that misses the flap is unusable.
Auditions are held against scene footage
Candidates read selected scenes with the picture running, which tests interpretation and timing simultaneously rather than in sequence.
Directors listen for a voice that sits correctly against the character's design and age, and for range across the emotional extremes the role will require.
Producers on the licensing side frequently approve final casting, and rights holders in Japan may have approval rights as well, which extends the process.
Recording is usually done alone
Most dub work is recorded with one performer at a time, following the picture and a guide track rather than acting opposite a scene partner.
That isolation demands a particular skill: constructing a performance from direction and timing cues without the reactions a live scene would provide.
It also makes scheduling flexible, which matters because a series may record over months while performers work on other projects.
Continuity keeps performers in roles for years
Audiences attach strongly to a voice, and replacing one in a long-running series is noticeable in a way that replacing a minor crew member is not.
Productions therefore work hard to retain performers across seasons, films and games associated with the same character.
When replacement is unavoidable, successors are often cast for similarity to the established voice rather than for their own interpretation, which is a real constraint on the performance.
Simulcast schedules compress everything
Dubs increasingly release close behind the subtitled episode, which leaves a short window for adaptation, casting approvals, recording and mixing.
The compression favors experienced performers who deliver usable takes quickly, which narrows the pool that regularly appears in fast-turnaround productions.
It also concentrates work in a few recording hubs, since a distributed cast cannot reliably meet a weekly deadline across time zones.