I saw that the latest beta has a new feature for creating a rough cut using the transcript. While at this early stage it seems a bit too clunky for my use case, I think the potential is there for a game-changer to workflow in the future.
Current State: With the current state of the feature (from my understanding), you manually select text from the transcript which effectively adds in-out points around that text. Then, you drag the clip onto the timeline via the project panel for example.
Idea: The software would use the transcript to predict "sections" of the same spoken text, such as a person doing a re-take of the same line multiple times. Then it could visually represent these sections in the timeline, and allow the user to choose which one to use or add to another sequence. Perhaps it could even take in a pre-made script and use that to better line up what sections go to what part of the final script.
Example Use Case & Concept Image:
For example, I usually have an outline of what I want to say in a video, but repeat each point multiple times before saying it in a way I'm satisfied with, and often stopping in the middle to start that point over. Meaning I need to go through every uncut video and find usually the last time I said a certain phrase and use that.
In the below concept I quickly made in photoshop, I'd imagine the software would highlight segments in the timeline that it detects are repeats, and color code them. So the color represents parts where I'm saying the same thing, but there are divisions for "takes" or times I restarted. Then the user could simply click on which one they want to use as the final take. Then repeat for each section.
And actually 8 years ago there was a feature demonstrated at Adobe MAX with a similar idea that was never implemented: https://www.youtube.com/watch?v=AvXWh-XB_hs . With the capabilities of today's AI transcription, I think such a feature would be even better than anyone could have imagined at that time.
4
Replies
AdChoices