Descript v79 adds per-corner radius controls for canvas layouts and expands AI Actions to auto-generate chapters and timestamps. The demo shows how to prompt the AI to create markers, plus fixes for SquadCast imports, timeline exports, and transcription prompts.
December 18, 2023. And earlier today, Descript rolled out version 79 of the software. And let's talk about what's new and what's changed. First of all, the big one is there's new canvas options. As far as being able to manipulate the canvas, move it around and how you size it. So the first thing is in the layout. If you have your layer selected, you could already change how round the corners were of your layer with this option in the bottom right under the layout section. And if you type a number in there, let's type in 50. Okay, so I typed in 50 and it rounded the corners a little bit. But let me get more dramatic. Let me type in 250. Okay. And look at that. You can see the corners are round. When you do this, it applies to all four corners of the layer that you're applying this to. What descript has added is the ability to control each corner individually. So right here, to the right of it, there's this little rectangle. If you click on that, you can see there's a top left, top right, bottom left. And because I set the overall one, they're all 250, but I can individually control these. So let's make the top left 25. Now, you can barely see that as round. I could take this one to zero. And, yeah, you can just play around with it and it just gets you more controls. And that's what descript is all about. Every update, you get more features, you get more options, and it's building towards a more fully featured editing software. That is cool. That's a cool little upgrade that we get. The next thing has to do with more AI action improvements. Descript has really been focusing on those lately. And the ones we have today are the ability to add chapters or markers, as well as the Add Timestamps option. And the AI action that we have added today is the Add Chapters or Markers, which is improved with a feature called Add Timestamps. So let's talk about it. If you go to your AI Actions button and the top of your transcript, and then you type in Add Chapters, it defaults there. It says Add Chapters or Markers, but you can also say, help me add timestamps. Help me add YouTube chapters. Just talk to it. It's just the AI, so it'll figure out what you need. But we click on that feature and this is the prompt that it automatically puts in there. Add Chapters to this script, aim for. And then you could input how many chapters you want. So I can just delete that and I Can say aim for four chapters overall. And I hit enter. AI is thinking about it. I'll just give it a moment. And there we go. So it came up with 1, 2, 3, 4 chapters. And so this is what's new about it is this add timestamps button right here. If I click on it, look at that. It gave me the timestamps for those places that identified that would make for a good chapter where the topic is changing enough. That is a unique topic within the video. And I can hit start over to re prompt it, I can hit close to get rid of it, or I can hit add to script to go ahead and apply that to this video. And there it is. Those are what it called the chapters, those bold parts there. And we have four of them. I had a couple that were in there already, like this marker one that I had already added, but that's pretty cool. And you can see how if you had an hour long podcast or something, you run that bad boy and get 20 chapters without having to review it and add those all yourself. Now of course you want to check them and make sure the quality of those markers is good, but it's one more step in the direction of all of this becoming automated. The next thing is they made some improvements to the speaker labels and the AI speakers functionality. So with the AI speaker labels it looks a little bit different. So if I click on this, there's has my name, the name of the AI speaker and then a. What do you call that? Like an exponent kind of looking thing with AI in the top right, a subscript. And so that just looks a little bit different. And then the words that are AI words, AI speaker words are blue on the transcript, whereas words that are from a human or just recorded words, whatever they are, are black like this. So that is one little difference with the UI is there's different colors and things just look, they're a little bit darker as well. The idea there is that it just stands out a little bit more. Additionally, when you click on one of these dropdowns, they split it up by ones that are in this project. So the, the speakers or the audio files that are in this project and then AI speakers that are not in this project are below that. And then of course you can browse your stock AI speakers and create a new one right there. And if you have any speakers in there that were trained on the legacy overdub before two months ago, or whenever they added AI speakers, then it'll say legacy AI instead of just AI. So that's it for the UI updates and the functionality updates. We also have some bug fixes and I'll go ahead and read these. We have a fix an issue where the end of videos imported from SquadCast would become resized that's been fixed. An issue with room tone clips not appearing in timeline exports color adjustments not applying to the script track filler words followed by a comma that would not get removed. Bug preventing copy and paste layout from updating the multicam selection within a scene. Always ask before transcribing. Still proceeding to transcribe the file even when the X was clicked. That's going to make a lot of people happy that we're hitting their limit on transcription and fix an issue where resizing the timeline prevents the full timeline from being drawn. All right, so as always, product's getting better and we're here for it. Talk to you later.
