Speech to text feedback on interviews

[Jules5C43]

Hello, 

Just a bit of feedback on the STT feature, coming from a video production team involving professional editors, subtitlers and production managers.

As a video production team, our work involves interviews of individuals that are facing the camera. We tested the TTS feature in Adobe Premiere Pro Beta and it looks very promising. On the other hand, there is certain features that we feel are missing and other relative issues concerning the already implemented features.

A bit of context before listing what we noticed : as of today, the most convenient way for us to subtitle our videos is to use a .mogrt file that we created ourselves through After Effects. We therefore have a standardized subtitle format that we can use by modifying the text only through the essential graphics panel. 

In so, with the STT implementation, we recreated our subtitle template by using the modifying features located in the essential graphics panel when clicking on a subtitle element and creating a style for it. The modifying features were enough to reproduce our template so we found it convenient to use.

The first issue we encountered was that while we are committed not to exceed 2 lines of subtitles for our videos, our style of subtitle is making it hard for the software to keep it under 2 lines, often placing the subtitles on 3 or 4 lines. 

The solution we found to this problem is to reduce the number of characters when creating a subtitle track, but we have no proper way of knowing what is the maximum amount of characters we can put in a subtitle element without ever exceeding the 2 lines limit. Another thing to note is that our .mogrt file for subtitles does allow more characters than 42, going around 60, without ever exceeding 2 lines.

Another important issue we found in the TTS feature is that we are used to place our subtitles precisely on the video cuts, as we do it manually. Our subtitles elements are synchronized with the cuts the video has. We think that it is more pleasant for the viewer and looks less glitchy in the overall look of the video. Obviously, as we may imagine, the subtitle track is reacting to sound, and not to video, therefore following the person's voice and not taking in account the scene changes.

Also, we have noticed little issues when transferring the Pr project such as subs appearing on the preview panel but not on the timeline, or visual modifications made to the subtitle style getting lost in the process.

I hope this has been a helpful read, feel free to ask any questions, I'll be happy to answer.

[Stan Jones]

Jules,

Staff have been less active here recently, but I know they value this type of input.

In my samples, I cannot get the S2T "Create Caption" step to produce more than 2 lines, no matter what settings I use. 

Are you applying the style in the "Create Caption" dialogue?

I do not understand the reasoning to limiting the maximum characters to 42. This is as if they designed that for closed captions, and we know from staff posts that they use the 42 characters as the limit for closed caption/wide screen. In any event, I completely agree that the options in that dialogue need to be even more flexible.

Stan