New captions - how do you import and use an existing transcript to create captions?

[JamesAOS]

1. What format must a provided caption be, in order to have it be importable to Premiere?

2. Once they are in their, is Adobe capable of analysing them and timing them to captions?

Often times we are using a teleptrompter and the script already exists. I want to be able to use this to get the right spelling for pro-nouns like people's names that Sensei doesn't get right. I also don't like having to rely on the cloud to do the analysis for me (only because it's been buggy and offline - right now my captions workflow is frozen again and my team are disgruntedly starting to type and time things the "old" way).

I'm assuming even if I successfully imported an existing transcript, Premiere would still need to connect to Sensei servers to match the text to the speech to time the captions, right?

Anyone else struggling to use existing transcripts?

[Alexander Riss]

Thank you for your feedback.

Currently the only ways to add a transcription to a sequence is either via transcribing, or via importing a previously exported prtranscript file.  There is no supported way to import a 3rd party transcript.

This also means that only Premiere Pro created transcripts can be run through the Sensei AI (running locally on your machine) to re-flow the text into caption segments.

On the offline problem - can you give some more details on this ? Are you on Premiere Pro 15.4 ? Behind a firewall ?

Best regards,

 Alexander

[JamesAOS]

No firewall - it was trying to do it with just the proxies and failing. Maybe a proxy/codec issue but reconnected media, it went through fine.

[Chuck CC]

A follow-up question to using the import transcript function in  Speech to text. The "import transcript" function is greyed out in my workspace. I am working on a Spanish captioning and have the word for word transcription in Spanish in docx format. Is there any way I can upload the transcription into the transcript/caption workspace to create captions?

[Stan Jones]

"Import transcript" is only available if you have no transcript already. But, no, you cannot import anything other than the PR transcript format (bascially what you would export).

You must instead focus on importing captions. In the Captions tab of the Text Panel, when there is no Caption track (or any existing caption tracks are disabled), you will see "Import captions from file." BUT the only file it allows so far is .srt. So you must convert your .docx to .srt - and that requires special formatting and timecodes.

See this post:

https://community.adobe.com/t5/premiere-pro-discussions/possible-to-import-docx-premiere-to-do-subti...

Let us know if you have questions.

Stan

[R Neil Haugen]

This is a bit of a problem currently ... as you can easily have a project with a transcript, that you need to import another language or additional material. The developers have been pushed about this by many users, so I expect at some point we'll get it.

But ... who knows when. Maybe, ... the Shadow knows?

😉

Neil

[DemSign]

What I usually end up doing is auto transcript the video in it's native language. Generate captions and export an SRT in the native language. Open the SRT in my favorite text editor, run it through a machine translator then have a native speaker edit my machine translation with time codes in place. Re-import the SRT with my now translated captions. 

[mbrazina]

Thats so bad. We are waiting for more supported languages like Czech for long time now. And now I see that there is no option to transcribe it by myself to make it work with Sensei AI. When there will be more supported languages?

[Stan Jones]

James, the new Speech to Text is a great addition. My experience has been that once you get the cloud issues sorted, it works consistently. And I only see the cloud activity for the transcription itself; not the caption creation step.

And yes, there are other users that want the same options. Submit your own feature request or upvote one that is already there.

https://adobe-video.uservoice.com/forums/911233-premiere-pro

If you submit a feature request, post a link here so we can upvote it.

The closest one I see is this one for you to upvote:

https://adobe-video.uservoice.com/forums/911233-premiere-pro/suggestions/43157127-could-we-use-adobe...

Other related requests you might want to upvote:

In the Speech-to-Text function, allow for the generated text/captions to be exported as a .doc or .txt file for proofreading

https://adobe-video.uservoice.com/forums/911233-premiere-pro/suggestions/42826278-in-the-speech-to-t...

Export Subtitles for Corrections

https://adobe-video.uservoice.com/forums/911233-premiere-pro/suggestions/42651067-export-subtitles-f...

Stan

[JamesAOS · Accepted solution]

Oh agreed it's absolutely fanastic (so much so it's instantly become a major loss without it) but I have an update. I tried a bunch of different projects and found that this was the only one failing to get past the audio render step (despitre the progress bar reaching what I see as the end). Then I checked and confirmed that the footage was actually offline. I'm looking at proxies so it's not noticeable and the new Text panel doesn't complain about the footage being offline. So... I reconnected it and, success, the auto-transcribe feature went all the way through and I was able to create captions. So... yay me!

It might be a good idea for Adobe to include a pop-up to say "hey this footage is actually offline". Perhaps it's trying to use the proxy audio and having issues with my proxies. I can export using proxies so it should work, maybe a bug. Anyway, if anyone else is getting slammed by this just make sure your footage is connected 🙂

[Rag and Bone]

Depending on how much work of this sort you're going to be doing,   Transcriptive might be worth checking out.

I haven't used it much, but they have just introduced a web version with a monthly subscription, and I think they have a free trial.

It can align an existing text with an audio track, and the resulting file can be used to produce captions: see here.

[R Neil Haugen]

I was happy to vote on the UserVoice postings that Stan listed.

Thanks for looking those up, Stan!

Neil

[leea66458980]

My workaround for this has been to let PR go ahead and do a transcription. It's been very easy to open my word doc and cut/paste blocks of text from the word doc into the transcript. Occasionally it causes the timing to drift a tiny bit but that seems to get fixed once I've been create captions from the transcript.

Hope that's helpful!

[tomatochip]

Adding my own workaround - 

After auto generating a transcript within Premiere - which is not even usable with proper nouns, company names, etc... - I was able to import a "corrected transcript" usign a plain text file from the original script. The auto generated transcript gave me the correct cuts/timecodes, and the "corrected transcript" provided the accurate text. 

[Stan Jones]

@tomatochip,

Thanks for posting. This is an old thread, and your workaround, made possible by the "import corrected" function, can work like magic!

Stan