Version 78 - New AI Features and Updates - Descript

Another day, another update. Today we have version 78. It's a bit unusual because this release is actually version 78, not 78.0 like it should be. They did that because they had some issues where they had to release 78.1 prematurely as a patch. So version 79, the next one, will be back to the normal naming standard, the normal naming convention. So with 78, what's new? Mostly, we have a lot of improvements to AI Actions, a couple of workflow things, and a whole bunch of bug fixes for the AI actions. They tell us that the backend processing has been improved so that it should be faster and a lot more stable. Also, when you receive an AI Actions response, it is now paginated. So before it was one big block of text. Now it breaks it into pages and maintains a history of the responses. Before, it only had one answer on the screen. Now it maintains a history of the answers so that you can page through them and review the prior responses. Let me show you. So here on my AI actions, I had it write two different things to expand on my project. And you can see there's now pagination, there's page one and page two. I never lose my history of previous prompts, previous prompt responses, and if I liked a previous one better than my latest one, I can simply go back and find that one and then add it to script, replace it, start over, et cetera. That is the new pagination in AI Actions. The next thing is that they have made improvements to the long script success rate. So before, if you had descript generate a script for you, the AI actions generate a script for you of anything longer than a few minutes, it was prone to errors. I mean, it's always prone to errors. This, this AI, it's very new, it's in beta, but they say that that is now greatly improved. When you have it generating scripts for you, it's going to have a lot higher success rate, completion rate, et cetera. And then lastly, on the AI Actions upgrades, they have made the AI Actions tools easier to navigate by breaking them down into categories. For example, let me click out of this one, close that. If I go back to my AI Actions, I go down to ask AI. Now they're broken down into the categories of Write, Edit, Find, and Publish. So right here you can see this header that says Edit. And then under it are all the different editing options that are available to you. And then there's find. You can find good clips, Find when. Under that there's Publish. You can have it summarize suggest title, make social posts, blog posts, et cetera, or write. You can have it rewrite things, shorten things, expand on things just one way that they made it a little bit easier to navigate and find things that are important. The next thing that we're going to talk about is the Correct all button. And this is actually really cool, really useful, especially with long projects. Like, for example, I see one word right here. This is a good, perfect example. The word organize. I have clients in the UK who use British spelling and they use the word organize with an S. And there was a way that I could replace all before, but now if I go to the correct menu, either by highlighting the word and hitting correct or holding C on my keyboard and then pressing the word, it brings up my correct menu. That's all the same. And then you still have your correct button, but now we also have a correct all. And this will correct every instance of that word. So it'll find every instance of the word organized with a Z. And then I'm going to change it to an S. And I don't know if I have any other ones in this video, but it would, in theory, correct all of them in one go. And I do have another one. It's telling me that there's two. And this was the old way to do it. This is what I used to do for clients. So just a little workflow, little thing that makes it a little bit easier. Now hit correct. All right, and done. Boom. And then next on the workflow stuff, they made it so that if you use the Add File button in a composition. So if I'm. If I'm here, I have something selected on the timeline. This Add File button, when I add a file, for example, another video file or audio, whatever it might be, it automatically adds it to your script and automatically starts transcribing it. So, fairly minor thing, but a nice little, nice little thing nonetheless. Moving on to just other updates and bug fixes. I'm just going to read these. These are off of feedback. Descript.com changelog. So the escape key no longer resets your tool selection. In previous versions, pressing Escape would return you to the default Select Tool. So if you're in Range Tool, you're in Blade Tool or anything else. Pressing Escape returned you to the select tool that no longer does that. They got rid of that. The next thing they tell us is the AI speaker authorization has been simplified so you can upload a consent statement in fewer steps. I just went through the process to see what they were talking about. With that, and the process appears to be the same. I'm not exactly sure what changed with that. It's still extremely easy. It only takes less than a minute to go from navigating to the AI speaker tab to having it trained, so it's already really easy. But apparently they've gotten rid of a couple steps Before Command W was the way that you entered write mode. You could press command +e or command +w both work. Now they have gotten rid of command +w and command +e is the only way that you can enter write mode going forward. Next is the Stock AI speakers can no longer be used to replace overdub prerecorded material, but they can still be used with AI speaker text entry. So you can't just replace words in the middle of your speech with stock overdub is what this is telling us. That makes sense. There's probably not very many situations where you would want to do that if you're, you know, it's your voice and then all of a sudden there's the stock voice randomly in there. It doesn't do that anymore, but you can still use it as the text to speech for an entire script or an entire section where you have labeled it with that stock speaker voice. Moving on to bug fixes Got a bunch here. Fix an issue causing Share page to loop at the end of playback. Fix an issue with clicking to move the cursor when renaming files. Fix the new composition keyboard shortcut which was not responding. Fix an issue allowing the clip speed value to change while using the Contact us form. Fix an issue with color adjustments not applying to the script track. Fix a rare issue causing an error when using remove filler words with wordless media present Fix an issue causing an error during playback message after uploading files Fix an issue preventing scenes from moving with an error generating text message. Fix an issue creating black bars around recording recorded video specifically on Logitech devices. It says so if you're using a Logitech webcam or something, maybe that was an issue you had and maybe it's fixed by now. We fixed several bugs related to the new wordless media in the script that should improve the reliability and performance of the editor, especially with longer projects. And fix an issue where placing a scene boundary marker and a speaker label change together caused the speaker label to vanish. And real quick, before I sign off, there's one other thing that wasn't mentioned in the changelog, but it's kind of significant, which is that the timeline looks different before our layers were color coded so everything was blue. That was a video layer, title caption, et cetera. And then audio layers were green. Now they look like this. They're all just gray. And it also creates a layer for the microphone that you recorded with your MacBook Air microphone or your external microphone. And they're kind of rounded. Before they were just the rectangular blue or green. So, interesting little UI change there. So the product continues to get better, and I'll continue to update you on it. Have a good one.