Plain text files in 5.0.0

[pandalamp]

On 4.0.15 and previous versions of Subtitle Edit, opening or importing plain text (TXT) files brought up a screen with a lot of options, including a toggle to "Generate time codes":

In 5.0.0 (beta10), plain text files can no longer be opened. They must be imported through the File > Import > Plain text... option, must have a TXT extension, and the window has less options. It is especially missing the "Generate time codes" toggle:

When working with plaintext in pre 5.0 versions, 'missing' time codes seemed translated to 99:59:59,999 --> 99:59:59,999, and could be manually synced via keypress using the Create/adjust shortcut: Key down=set start, Key up=set end and go to next. 5.0.0 does have a General shortcut to: Insert subtitle: Start on key down, set end on key up, but that option only creates a new line.

In my workflow at least, I transcribe to text files breaking the lines how I want them, then import that file into Subtitle Edit and manually sync each line as I watch the video. This doesn't seem possible yet in SE5.

[darnn]

I second all of this. The one thing I can offer is that using these settings:

You will get, as you can probably guess from the image, a subtitle file like this:

se5textimportsample.txt

And you can now sync it using these keyboard shortcuts: Set start, Set end, Set end and start of next plus gap and go to next.

[darnn]

To avoid doing all of this manually, you can also use Whisper and WhisperTimeSync to create a subtitle file with somewhat-accurate timing and then combine it with your accurate transcript.

For example, take this trailer:
https://www.youtube.com/watch?v=Ams0ROhWX0w
Here is an accurate transcript:

merrily.txt

I've used Whisper - Video -> Speech to text, engine: Whisper CPP, model: large-v3-turbo - to create a subtitle file:

merrily-whisper.srt.txt

And then WhisperTimeSync:
https://github.com/EtienneAb3d/WhisperTimeSync/
With this command:
"C:\WhisperTimeSync\distrib\WhisperTimeSync.jar" "D:\merrily-whisper.srt" "D:\merrily.txt" en

And this is the resulting file:

merrily.txt.srt.txt

As you can see, it's not perfect. You need to move the first word into the first subtitle, and the jar file only handles about 800 subtitles per file, so you have to break larger files up into chunks.

Compare this with an accurate subtitle file I made:

merrily-mine.srt.txt

I'd say the one with Whisper timecodes is a much better starting point than blank timecodes. You can achieve even better results using e.g. the large-v2 model.

[pandalamp]

To avoid doing all of this manually, you can also use Whisper and WhisperTimeSync to create a subtitle file with somewhat-accurate timing and then combine it with your accurate transcript.

The thing is, I like doing it manually. It feels really satisfying. Getting it right, the way I want it the first time through, is why I am writing and timing subtitles from scratch. Thank you both for posting workarounds and examples though.

My intent in posting this issue is to say: Hey! Somebody actually uses that checkbox and that shortcut :D

[darnn]

Truth be told, when I care about accuracy, it's rare that I leave a single timecode the way I got it from Whisper. But it's still useful to have. Plus, sometimes I don't care about accuracy, e.g. if someone else is timing it after me.