You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It'd be useful to be able to make adjustments to an existing transcription which is not fully accurate or incomplete.
A rough outline of the feature is that Parlatype would first read a list of tuples of the form (start_time, stop_time, text/identifier). For simplicity, let's say these are stored in tab-separated lines:
0:00:00.000 0:00:01.234 Hello
0:00:01.234 0:00:03.569 Hi, how are you?
0:00:03.569 0:00:05.189 That's enough…
Upon reading this input file, Parlatype would then display these non-overlapping sections in the waveform area – as if having many selections, perhaps differentiated by color. The user could then adjust the timing of each section, using the keyboard or dragging with the mouse. The text/id of each section would also be displayed (maybe in a separate area) to make clear which section is which. The modified version can then be saved.
I hope this feature seems reasonable if you have a chance to take a look sometime.
The text was updated successfully, but these errors were encountered:
Thanks for your report and sorry for the really late reply! I've been aware of your report and it sounds very sensible. The problem is that Parlatype doesn't have its own editor window, mainly because I want to avoid the maintenance burden. At the moment it's too difficult for me to implement this. It would need a lot of bidirectional communication between Parlatype and LibreOffice.
I leave this issue open as I think it would be an improvement, however don't expect a solution anytime soon.
No problem at all about the late reply. I actually think the feature is simpler than you describe, so perhaps I can explain in another way:
You already have the concept of a selection. This has a start and end time, and the user can adjust these times with visual feedback. So, fundamentally, the feature I propose is that instead of just one selection, there can be an array of selections.
A menu item "Load" reads the selections into an array from a file in, say, csv format (id, start, stop)
The start/stop time of each selection in the selection array can be adjusted by the user
There is a restriction that the selections do not overlap
The selection struct has an extra field id (or similar) which stores the selection's identifier or text.
This id/text is shown as a tooltip or in another text box
A menu item "Save" writes back the selections to disk
So, it seems to me to be only an extension to the existing user interface (showing more selections), certainly nothing that would require a separate editor window or LibreOffice (which I don't even use).
Anyway, this is just an idea for your project that I think would be useful to people, so no problem if you decide not to go ahead, and thanks for the reply.
Hello and thank you for this program.
It'd be useful to be able to make adjustments to an existing transcription which is not fully accurate or incomplete.
A rough outline of the feature is that Parlatype would first read a list of tuples of the form
(start_time, stop_time, text/identifier)
. For simplicity, let's say these are stored in tab-separated lines:Upon reading this input file, Parlatype would then display these non-overlapping sections in the waveform area – as if having many selections, perhaps differentiated by color. The user could then adjust the timing of each section, using the keyboard or dragging with the mouse. The text/id of each section would also be displayed (maybe in a separate area) to make clear which section is which. The modified version can then be saved.
I hope this feature seems reasonable if you have a chance to take a look sometime.
The text was updated successfully, but these errors were encountered: