Closed xhzhu0628 closed 3 months ago
Thanks, my issue has always been overthinking it and trying to get queueing to also work with ASR/Detect-language.
My limited testing shows it following the queueing nicely.
Take a look and let me know what you think. I added a check under gen_subtitles_queue
so it won't queue something unless it actually needs to work on it.
If CONCURRENT_TRANSCRIPTIONS
is 1, it will continually re-create the model on completion (not a huge deal, just a second wasted reloading the model and spamming the log).
Also, the Fixed this in latestfiles_to_transcribe
needs to continue to exist in its current instantiation because that's what is used to 'internally' track the files being worked on so we don't accidentally delete/purge the model while we're working on it (primarily from the ASR and detect-language side because they don't call gen_subtitles
)
If we can figure out how to queue up the ASR/Detect-language as well, that would be great (or at least be aware when they are running).
Sorry for spamming your branch, trying to make edits on my phone.
fix queuing .#51