Hello yall! Love the new feature for transcription. Honestly have to say that's the main reason why I haven't fully switched to Resolve. What is the best practice for transcribing many hours of footage? I have almost a terabyte of video files that I want to transcribe but I know that is a lot to do at once. Any tips?
My first thought was to focus on the parts that come after the actual transcription. How are the captions created and can you save time by looking at the settings available? (Some users report success and others can find no way to get the captions to come in correctly.) Can you use export of the transcript to make editing easier? (Not really a workable function with the current options, but a popular request.)
This depends on your purpose. For an archival database I might not care about accuracy or even accurate timecodes (you can't currently export the transcript with timecodes). I would still have a much better resource for looking for material.
I had assumed the main task, creating many transcriptions, had to be brute force - you can't queue up a bunch of sequences for transcription in Media Encoder. But make a feature request (or find ones that you like) and post a link(s) here for us to upvote:
Then I did a little test, and discovered that you can have more than one sequence being transcribed at a time. I have no idea what the limits are. Does Adobe limit the number of transcriptions you can do simultaneously? Inquiring minds want to know!
I was running a 25 second clip in separate sequences. I wonder if there's a strategy to the computing demands on the computer vs the upload times. I do not know, but looking at the progress display, I think the steps are "rendeering" the audio (computer based), uploading the audio (internet connection based), processing the actual transcription (occurs on Adobe server I assume), and then downloading the results (I don't know if this occurs throughout the process or at the end).
For a very simple, two and a half minute sequence, the rendering took a few seconds, the upload was very fast, the processing took a minute or two, and the appearance of the transcription in the window seemed instantaneous at the end of the processing step - as determined by the progress window. In any event, the bulk of the process is in the cloud. You don't need to automate the "setting up" part - create a sequence, start the process, and start setting up another one. For long sequences, the waiting time will be the issue. But having multiple sequences running overnight would make the whole process shorter.
Let's see what others think, and let us know what you find.
Thank you Stan! Very thoughtful answer. last night I tried to transcribe a 10-hour sequence overnight and it froze up. So if I do many hour or two sequences that might work I think?
To answer your question about my goal, I want to be able to search the text and also be able to jump quickly to the part of the video where they say it. I was able to do this easily with just a few hours, but will each sequence be searchable? What if after I transcribe, I move all the files to one single sequence? Would that cause a problem or even make sense? I will have to play around to see what happens.