1) Export script from ChatGPT as plain text/Markdown; keep scene markers and timestamps.
2) Convert to SSML: wrap sentences with , add for pauses, use or similar for consistent pacing, and validate SSML.
3) Batch TTS in ElevenLabs: create and save a voice preset, upload SSML files or call the API in batches, keep the same sample rate/preset for every episode, and name files with episode numbers. Test one episode first.