I have a scheduling task and would like to ask everyone for some suggestions

I have a long verbatim transcript, possibly exceeding 50,000 words. I’ve written a program to split each person’s dialogue into BeforeTranscriptSegment. However, I have several issues, which I’ve listed below. Please provide me with some suggestions. Thank you.

By default, I use Gemini for translation because its input token count and output token count meet my expectations.

  • The Gemini plan I’m using has a RateLimit, so I want the API request for translation to - trigger only once every 4 seconds.
  • If there is no transcript to translate at the moment, I’d also like the 4-second worker not to trigger.
  • If there are untranslated segments or if an API call returns an error, I want to implement a retry mechanism. Could you suggest a good way to design this task? Thank you.
Вернуться на верх