Describe a character and generate a sample
Begin with the role, then describe timbre, language and delivery in plain words. A warm narrator with measured pacing is easier to evaluate than a long list of unrelated qualities. Add the emotional distance you want: conversational, formal, intimate or reserved. Make the description specific to your original character rather than naming a real person to imitate.
Choose a short preview passage that tests the sounds and feelings important to your project. For an explanatory host, include a question and a concise answer. For a dramatic narrator, include a change of mood. Listen to the saved sample before casting the character. If the direction misses the mark, refine the description and compare a new sample. Reuse a similar audition passage when judging alternatives, so differences in the writing do not make the voice comparison harder.
Cast recurring characters with clear names
Save a distinct name for each synthetic character and use that name consistently in dialogue. Select up to three ready saved characters for a segment. Think about contrast as well as individual sound: two voices may be pleasant on their own yet hard to tell apart in a conversation. Preview the actual exchange and adjust the casting or directions when listeners could lose track.
Keep voice references close to the script
Use the voice library as a starting point for a project’s cast, then review each generated segment. A saved synthetic sample helps provide a reference; it does not remove the need to check pronunciation, performance and speaker identity in a new scene. If you change a segment’s cast after generating, its existing audio becomes stale. Generate a current version before exporting that revised scene.