Transana looks interesting for analyzing videos and PDFs with notes and categorization

[jsn]

Continuing the discussion from Qualitative Research - starting new DTP2 DB, suggestions?:

@rabourn

I downloaded Transana and was quickly frustrated. It’s difficult to try a product when it’s crippleware. I could never be sure if I didn’t understand it or if a feature was missing or intentionally disabled. I don’t think disabling features is a good way to let people explore your software. They would have to already understand how it works to figure out what it can do for them.

Transana.com seems to offer similar AI capabilities to DT4 where we can bring our own key for ChatGPT.

The software trial is crippled in terms of size of transcript editing and AI, but they offer a way to pay month by month for $25, so that seems reasonable. For somebody just doing video transcription, even Descript.com charges that much for their minimal tier to get a transcript that is tied word by word to the video and as far as I can see, Transana gives a lot more features on the text-side than Descript, which is more focused on republishing and fancy video editing despite giving you a word-document experience for a transcript.

What I find interesting about Transana is that it keeps everything local, like DT, and has database capabilities such as hierarchy of tags.

Their youtube channel has a lot of videos showing how it works. Unless you guys find something better, this looks like a winner to me. Anybody else agree/disagree and have alternatives that are clearly a better value?

Let me explain what I need and what I like

• I need to be able to see the video in high resolution because there are details I want to note in the transcript.

• I need a transcript that allows me to type notes along with the text and Descript does this with their () which they call inline notes.

• I need to be able to export the transcript with the notes and markers because I need to put it into a database where I can categorize it and tag it and ask LLM questions. I use DEVONthink for this, but if the tools are integrated and no export is needed, that is great as well. For example, if I can easily find clips from all videos that are associated with a specific detail I have categorized, that is ideal.

• I like to add markers as titles so I can skip to certain areas of the video and review.

• I like to be able to click on a word and be shown that precise moment in the video.

• I like to be able to draw on the video or insert small screenshots into the video over a certain time period to be used as notes.

Apparently, Transana has a built in screenshot tool so you can take annotations right in the transcript (although I don’t see a way to edit them) and it lets you "clip” sections of video and mark them with tags in a hierarchy with notes. This is similar to the way DT allows us to link to a video frame, but storing the whole clip reference is obviously very interesting for compiling a specific analysis across a huge library of videos.

One thing I’m not liking is that it doesn’t feel like a mac app. It doesn’t use finder for locating files, for example.

[jsn]

This is pretty nifty. Multiple timecoded simultaneous transcripts.

That is very nice as it allows you to make your own annotations about what is happening in the video while not disturbing the audio transcript. So, you can note gestures or technical details about a model that are not obvious from the audio for that specific frame of the video. The fact that each of those annotations can also be coded with information such as a keyword and/or added to a clip with title and description and stored in a hierarchy of properties makes this feel exactly like DEVONthink, but for videos. Really amazing to me having always hoped that DT4 might do something like this for Video like they have done with PDF - i.e. giving me a way to look at the sounds, objects, gestures, and other properties inside the video as if they were text (a video OCR, so to speak) and aligning it with the audio and video time codes.

I would like you to let me ask my question

Also explained in their tutorials: Advanced Time Codes | Transana Online Tutorial

[jsn]

The tool works well for my purposes, but I have had to setup some Keyboard Maestro replacements for some actions.

The developer (who seems to be a PC, rather than a Mac guy) chose to use cmd+A for his go back 10 seconds an play function. His goal was to keep his left hand on the keyboard and use cmd A,S,D,F to move through the video.

That being the case, if you want to have a select All function (typical for CMD+A both on PC and Mac) then you have to build it yourself. I did this:

The other problem I had was that I can’t stay in edit mode and also click around the time markers in the transcript. When switching from edit to read the program prompts you if you want to save and the dialog doesn’t support keyboard to save, although escape does cancel in the sense that it doesn’t save and causes loss of work. It should offer

• Escape - do nothing
• Enter - save

instead it does:

• Escape - do not save

Plus it’s a blocking dialog, so if you want to make an edit and then click over to another location in the transcript to play it automatically and then make another edit, you cannot do so without clicking the Edit/Read toggle button (no hotkey) and then clicking back in the transcript (moving your mouse again) and waiting for the playback of the video and then clicking again on the Edit/Read toggle to make changes to the transcript.

So, to handle all of that, I wrote a more complex KM macro that looks at the screen for the icon state, saves my work if it is in edit mode so that the dialog doesn’t show up, then switches the mode and puts the cursor back in the text where I typed, this allows the cmd+D to play automatically so I can listen without hitting it and then I can just type. So the workflow becomes:

• Hover with mouse over text
• Cmd+R to force read mode, play the current segment, return to edit mode and put the cursor where it was hovered
• Type the correction

This is fast and allows both sequential corrections and randomly clicking around a transcript to just correct the main issues or edit prior transcripts that need minor corrections quickly.

Another problem I have is that whenever I open a new transcript, the windows all resize. This resize operation also, if done while transcribing, causes the video to cursor to be moved to the beginning, so that means scrubbing back to your position again if you ever need to move a window. Partial solution is to uncheck: Options > Auto-Arrange (does not persist across sessions)

Just like the windows do not remember their position, the waveform doesn’t seem to have a permanent zoom either. I might develop a KM macro that will size the windows or try with Moom and see if it can keep a memory there, assuming the developer doesn’t answer my requests and issue reports.

There are problems with paste not being allowed (for me, but the docs say it should work: https://www.transana.com/tutorial/transcription-shortcut-keys/ and the PDF shortcut key list is different than the web version: https://www.transana.com/wp-content/uploads/2018/11/Transana-Keyboard-Shortcuts_2018.pdf) , so tools like KM that use paste to complete text must be changed to type instead and tools like FixKey which I use for AI requests like formatting dates and grammar and translations fail to work without a shim via KM that takes the clipboard results and types it into Transana:

I have also seen other issues like cursor disappearing, windows not closing and Force Quit being necessary, but I haven’t had a crash yet and I haven’t lost any work.

Great tool, but having to build workarounds to the UI may not be the favorite thing for most people.

That said, I don’t think I have any tools that actually let me do what I want. Even DT4 doesn’t let me access that little Script menu and I still feel it should just let me paste or use keywords for scripts in a chat window and work with the AI to work out bugs if necessary.

[jsn]

Discussion seems to have been hidden by the admin, so perhaps nobody will see this…

Digging deeper, I think I will not use this tool.

There is a better option:

What is nice about Transcriptions is that it appears to be a native mac app built in xcode and it just works out of the box. From DEVONthink, you just open your RTF file and then pick the media that goes with it and it just works. Apart from needing to switch the time coding to their format, which is easy to do, you can click on anything and edit it and it saves and DEVONthink and the mac just see it. So it’s a great way to have a transcript editor that is integrated with DEVONthink and no proprietary engines to learn since you can already do all the coding with the tools inside DEVONthink.

Release Transcriptions 1.4 · Transcriptions/Transcriptions

The problem with Transana and other tools like it is that they use proprietary databases.

I tried some things like taking data out of Descript and putting it into Transana.

Transana supports DOCX, but the way Descript writes the DOCX file, the highlighting is lost. Not surprising because DEVONthink also can’t read the highlighting.

RTF from descript goes into Transana but then when you export, it fails to produce the highlighting in the RTF it creates.

It’s not a native mac app so it has all sorts of other issues as well.

Frustrating. DEVONthink is a much better value since there is an open source option for editing the transcripts in situ.