Is your feature request related to a problem? Please describe.
I've experienced that when requesting a youtube video transcript in a language that requires translations, many times it requires a few attempts before such translation is downloadable; but most importantly, the session used (especially when proxying from a service account perspective at AWS for example) keeps being the same on each attempt towards the downloading process - resulting sometimes in blocked requests.
Describe the solution you'd like
Would it be possible (not sure) to generate a new session for each new attempt regardless of proxying or not?
Describe alternatives you've considered
Not considered alternatives, since I saw in the code that the translatable transcript object is the singleton object used for then performing the get request.
Additional context
Attempting to reduce the errored requests to make it closer to 0