Text Panel Enhancements: Share a transcript with a colleague for correction

[TeresaDemel]

Many of you have told us that you have a colleague responsible for correcting AI transcripts, and that you want an easy way to share the transcript and get corrections from them. Premiere Pro (Beta) now allows you to send the transcript to your copyeditor to make sure that brand names, lingo, names and titles are properly spelled and capitalized.

Watch the video below to see this workflow:

• Open up the Text Panel and make sure you are viewing the Source transcript
• In the Overflow menu select Export -> Export to text file
• Save the text file and send it to your copyeditor for corrections
• When you receive the corrected file, save it to your desktop 
• In the Overflow menu select Import -> Import corrected transcript
• Now you can use your corrected Source transcript to create a transcribed sequence and generate captions

Try it out and let us know how it goes.

https://www.youtube.com/watch?v=pYmdN-nvQkQ

[jasonl22319738]

This feature seems to have a lot of bugs. My assistant corrected the transcript (not in the text file but in a separate project). I then imported those corrected clips into the original project, then exported their corrected transcrtipts as text files and then I imported them back onto the original clips in my project. (I couldn't simply erase the old clips and replace them with new ones because the old clips are being used in various sequences.) When viewing the "corrected" transcript there are now MUCH wider gaps between speakers (you can see that happen in the video you posted above when the new transcript is imported) Also, the speaker names have reverted back to defaults (speaker 1, speaker 2, etc) rather than the corrected speaker names in the new transcript (even though they appear correct when looking directly at the text file) Also breaks between different speakers are often missing. What I assumed would be a helpful time saver has wound up turning into a situation where I have to do the same job twice. Please fix this.

[Stan Jones]

@jasonl22319738,

Last issue first: speaker names cannot be edited via the "import corrected transcript." That would be a feature request. Definitely a good idea, but just not the way it currently works.

When this feature was introduced in Beta (3/3/23; original post above), it was clear that this is a limited "correction." And my tests in the release version (I think 23.4.0) and later Betas is that you cannot make major corrections. I have not confirmed this in the latest Betas. When you make corrections in a version other that the actual export .txt, the import may not work well.

You refer to importing clips. I'm not sure what you are doing. Using this feature, you can only import a corrected transcript to a source transcription or to a sequence, static transcript, depending on what you exported.

I cannot find a link to a method I posted attempting to import an external transcript.

Describe your process a bit more and I'll see if I can help.

Stan

[jasonl22319738]

I'm a trailer editor and this is my process: At the start of a new job we receive six to seven 20 minute clips. These are reels from an already cut feature film so there are a lot of different speakers and there are cuts imbedded in the footage (we don't have access to the original sequences used to create those reels) I import those clips into a new project. I assemble the all of the reels/cips into one long timeline - a "feature timeline" (so I can scrub and view the feature in one piece and not have to constantly remember which reel/clip something happened in) Then I use Scene Edit Detection to add cut marks at each detected cut. Then I use a pancake editing technique to pull good/useful shots (just the video, not audio) from the feature timeline into new selects timelines. The next step is to pull dialgue selects as subclips. A good transcription makes this easy - copy the text from the transcription, highlight the text (in and out marks automatically!!), create the subclip, and paste the text. No typing out of the dialogue needed. But..the auto transcripts have countless flaws - speaker names need to be added, dialogue corrected, multiple speakers in certain sections need to be split, the same speaker over multiple sections need to be merged, etc. This is a very time consuming process that can take me several days to complete - time I don't have. So while I am busy pulling the video selects, multiple assistants load the same feature reels into new projects and in Premiere, they correct the transcripts and place the corrected clips into a new project. Now I have my project with the first instance of those clips cut into various timelines (the feature timeline and my selects timelines) and I have a separate assistants' project with those same source clips but with a corrected transcript. The problem is - how do I get those corrected transcripts into my project? I can copy and paste those correct clips into my project sure. But then I have two instances of the same source clip - one correct, one not. And my feature timeline (now with thousands of add edits at each cut) is still linked to the incorrect clips. I tried exporting a .txt file of the correct transcript and then importing it back onto the incorrect clips but as discribed earlier it doesn't work properly - there are increased gaps in the interface between speakers. It's often missing the switch from one speaker to another (typically when the switch happens quickly like one word from speaker 1, then one word from speaker 2 then right back to speaker 1, etc), and it doesn't carry over speaker names despite them being correct in the .txt file. Trying to replace the old incorrect clips with the new ones is way too time consuming - remember there are now thousand of add edits in my feature timeline. It would seem like exporting as a .prtranscript would be useful in this situation but that format oddly can't be read back by Premiere (I'm not sure what use there is for that format at all because I can't find anything able to read it)

[Stan Jones]

@jasonl22319738,

Excellent post. And the Beta forum is a good place for it whether you are using the Beta version or not. Staff are actively working on these features, and real life workflows are very useful. Clearly excellent tools for your workflow, but the devil is in the details.

@TeresaDemel FYI. Any thoughts?

I have limited time at the moment, but let me respond to a few points.

So far, I think "import corrected" only works when you are importing that transcript back to its source. When the assistants create a new project, you lose the source transcripts, and the "corrected" transcript becomes an "external" transcript - i.e. as if you had your own transcript and want to import that. I would experiment with the assistants not changing anything but the transcription, exporting that as .txt. Then you importing that as corrected.

It would help to know what PR version you are using. Also, at what stage are you creating the transcript?

You refer to subclips; are these actual subclips as created by PR? And how do these relate to your major timeline?

I'm still trying to sort out export and import or the proprietary .prtranscript. If you are in PR 23.4 or later (text-based editing - source and sequence transcripts), import appears to only work for a static transcript and therefore only available when in sequence view. But you can export one from a source media transcript. Since you can't edit these, I have thought they are not useful for exchanging changes.

Stan

[jasonl22319738]

So far, I think "import corrected" only works when you are importing that transcript back to its source. When the assistants create a new project, you lose the source transcripts, and the "corrected" transcript becomes an "external" transcript - i.e. as if you had your own transcript and want to import that. I would experiment with the assistants not changing anything but the transcription, exporting that as .txt. Then you importing that as corrected.

I'm not sure how that would be possible. If import a clip and then it gets an auto transcript attached to it are you suggesting the assistant only edit the transcription in my project (rather than pulling the same source file into a separate project)? To do that, they'd have to do it in my project which I'm using. Also once the correct transcription is completed we will want to share it with multiple other editors who are on the same job and who all have their own separate projects at various stages of breaking down the source material. 

It would help to know what PR version you are using. Also, at what stage are you creating the transcript?

I'm using version 23.5.0 (build 56)

We want to transcribe at various stages throughout our process as new material comes in. But we want to start every job with a simultaneous breakdown by an editor or editors (ie creating subclips and selects sequences in their own projects) and a transcription correction by the assistants done on a separate system that can then be shared with all editors and will update the transcriptions in their projects.

You refer to subclips; are these actual subclips as created by PR? And how do these relate to your major timeline?

Yes. The subclips are created by Premiere.

At the end of my breakdown process I basically have 3 different types of assets to work with.

A single sequence with all of the feature reels (the original source clips) assembled in order with add edits at every point where there is a detected cut in the material.

Multiple selects sequences that have primarily video elements assembled according to a particular theme (For example, I might have a timeline that is only shots of a particular actor, or scope exterior shots, etc)

And finally I have subclips that are pulled from the original source clips that contain important/relevant/interesting dialogue lines with the name of the subclip changed to reflect the dialogue it contains.

Being able to copy bits of text from a correct transcript and paste them into the name field of a subclip has been a HUGE time saver as opposed to having to type out the dialogue (like we did before transcription was available).

[jasonl22319738]

As far as I can tell here is the issue and for me it's a big one. There is currently no way to import a transcript .txt file onto a clip that hasn't already been transcribed. If you transcribe that clip, PR decides where to break between speakers and sets in and out "marks" next that speaker's name. When that clip is corrected in a separate project and new breaks between speakers are created (or multiple instances of the same speaker are merged into one text section) and then that transcription is exported to a .txt file all of these new breaks are faithfully exported to the .txt file. BUT when you import that corrected .txt file back onto the original clip, PR forces the corrections into where the old breaks occured rather than taking the new timecode numbers from the corrected .txt file. Plus it ignores all changes to speakers names and reverts to generic names: Speaker 1, Speaker 2, etc. This is a major bug with regard to the way we hope to use the transcription feature. There should be a way to import a .txt transcript onto a source file without having to first generate a new (and inherently incorrect) transcription. All of the relevant data is in the .txt file. Why can't I properly access it within PR?

[adrichar]

can you export transcrips from a group of clips at once?

[jarrett_b]

Thanks for adding this feature. After importing corrected transcripts and creating captions, I'm experiencing a bug where occasionally there are "X" symbols appearing at the beginning or end of the caption, but only in the Program window. Once I double click the caption in the Captions window (which appears normal), the X symbols disappear. It's a minor nuisance, but any chance someone could take a look?

[dhwinston]

This exact situation is occuring for me. Our workflow for captions includes auto transcibing a final version of the program (output ProRes imported back into PR). Exporting that transcript to .TXT for proofreading (or comparison to Whisper export). Importing the corrected transcript and generating captions.

At the end of every one of the transcript's "paragraphs" 2 extra, empty lines are added that causes these empty characters to appear on the corresponding caption. Yesterday I just deleted the empty characters directly in the program viewer. Double clicking inside the caption in the captions editor does make them disappear, but then adds additional lines inside the caption that need to be deleted.

I managed to get rid of 1 of the empty characters by going back to the .TXT file and removing all empty lines between each paragraph and the timecode line. However, I still see one empty character on the caption that corresponds with the end of a transcript paragraph.

[dhwinston]

I was able to get both empty characters to go away by importing the corrected transcript as an unbroken wall of text. No carriage returns. No timecodes. No speaker names. PR brought the text in and everything still lined up. Good to know, but it would be great to have that adjusted.

[Stan Jones]

@dhwinston, Good job.

I reported this bug a while back:

https://community.adobe.com/t5/premiere-pro-beta-bugs/import-corrected-transcript-adds-spaces/idi-p/...

I can't find the thread, but staff reported that you could now do a merge all in the transcript. But if you did you lost the timecodes. So I tested what you did - merge all outside PR, then import corrected, and the times work. I used a single timecode for the whole range and one speaker.

Stan

[AndrewTheGreat]

Really cannot understand the decision to make the import of corrected transcript a Source clip only. I've got 28 takes in the current project and an edited video with a finalized transcript. Editing (actually translating the original language into my native one) in Premiere Pro via the Text panel is WAY awkward and inconvenient. I could export the edited transcript into a text file and import the translation back via the import trancsript menu but this option is not available in the project monitor. Who is that brain in Adobe who made it work in such a way?

[Stan Jones]

@AndrewTheGreat,

You CAN import corrected for a Static Transcript of a sequence. Just be sure you created one, and that you are in Sequence view in the Text panel/Transcript tab.

If you don't see that, post a screenshot of the timeline, Program Monitor, and Text Panel.

Stan

[AndrewTheGreat]

Thks for that. It works.

[Stan Jones]

Excellent! It really is a great feature.

Stan