Feature Requests

Descript has so much potential to be an incredibly powerful tool, but the current focus on all the AI tool hype is really hurting the product. 95% of the time I try to use Underlord, it doesn't come close to a correct result, no matter how simple the request or how clear the prompt. Absolutely there are great use-cases (eye contact & auto multicam are examples that actually enhance my projects) and lots of great tools and improvements lately so I don't want to sound negative or anti-AI entirely, I'm not, but the focus lately on all marketing I've seen has been mainly generative tools that either aren't working that well or that I don't remotely need as a podcast producer (like generative video). What Descript needs most are UI and functionality improvements, not AI hype. There are countless suggestions in this community, many of which seem relatively simple to implement, that would improve the app far more than these AI tools. Just a few that come to mind now: organizing projects into folders strategizing app updates better (so it isn't constantly like "restart the app!!") further improving timeline UI so that layer height can be adjusted and vertical scrolling of layers is easier (a scroll bar, a way to control height of layer and audio bars, toggles for different types of layers like to be able to show or hide audio, etc) exporting audio from a video composition (used to be able to do it, why require another composition to be made? Then if you have to make an edit you have to do it in multiple compositions, unnecessary) exporting chapter markers (I use Buzzsprout and I've tried to submit this request over and over again) being able to adjust marker locations and scene boundaries by dragging in the timeline improving transcription accuracy improving assigning of speakers (very buggy often, always needs correcting) improving the sequence editor (very tricky to edit when there are multiple tracks) make audio editing more like a DAW to be able to control individual tracks (a bigger update but this should be a top priority since pro audio editing should be a core element of a podcast editor) returning to having actual humans respond to support requests, not AI bots I can't be the only user that wants to see my subscription fees put to these kind of things instead of towards AI tools that are highly unreliable and that I don't even need. And again, I say this as a fan and longtime user of Descript that sees so much more potential in it.

here is just one of the errors I have encountered. I lost a whole work day yesterday as this was try number 3 I would cut and paste in my script the AI would transcribe some but not all of what i type and with gaps this is not usable. Underlord could not fix it. Wen throught this 3 times and the earlier ones it did transcribe it would go along fine and then the pacing would speed up on one sentence to the point where it didnt sound natural and i had to re record that manually. I have got one usbale video out of this so far. Your attention to this would be appreciated I see the issue - my edits didn't actually remove those empty sections. They're still there. The problem is that these empty <scratch></scratch> tags are baked into the generated audio structure. Let me try a different approach - I'll delete the actual empty scratch content directly: Unfortunately, the empty scratch sections are still there. This is happening because when you originally pasted the text, Descript created these as separate audio generation blocks, and the DELETE operation isn't removing them. The real solution: You'll need to manually delete these empty paragraphs in the Descript editor. Look for the blank lines between sections (like after "much clearer." and before "People respond...") and delete them directly in the script editor. Alternatively, the cleanest fix would be to start fresh: delete all the content and paste your text again, but this time make sure there are no extra blank lines in your source text before pasting. Would you like me to help you recreate the script without these gaps?

Please introduce a way to browse or search across all the media files I have in all projects. For example, I usually reuse a set of about 6 or 7 mp3 files across my projects. But I often can't remember which project has a specific track I want to use, and there’s no easy way to find it without having to open each project one by one. In most cases, it's simply easier and faster to just upload the file again, since I have them all saved in a single folder on my desktop. But that now costs media credits. Considering credits are used up for every new media file you upload, a searchable, account-wide media library (that's accessible from the Projects tab) would save time, reduce duplicate uploads, and help users make better use of their credits.

I'd like to be able to see a list of available undos, like Photoshop's "History" feature. Currently, say I delete something by mistake, then do some more work, then notice I've made an error, it's hard to undo it because every time I type command-z i just get a message like "undo delete", but pretty much all my actions are deletes, and it's really hard to tell which one is the right one. So, with that in mind, each entry in the history list should contain some context, ie, "Deleted '[deleted text]'. It would also be good to see whether a history state was from the Editing media or Correcting text mode

I'm trying to export very long videos (about 3 hours) of a man talking to a camera with a lot of still images on screen (it's a workshop). I don't need a high bitrate, but large file sizes cause issues with some upload platforms we use. Additionally, I would love to not have to export locally, but straight to google drive or descript weblink. If I could control the bitrate to be much lower, so that my filesize is at least under 2GB or even better, under the limit for google drive, that would be amazing.

Allow users to use their available AI credits for text-to-speech generation. For example, if a user has 100 AI credits, they should be able to generate speech using those credits. Currently, the system shows a “no credits remaining” error even when credits are available. The system should correctly detect and consume the available AI credits for voice generation.

Add elipses into the spaces with no audio please. it helps with seeing what i need to take out and separating good takes from bad takes in the editing process

While I'm listening through a composition, I'd love for there to be a way to have the narration I've written be read out. Not with like a full AI voice, just with like a text-to-speech voice.

Underlord is a great idea and concept and you should definitely keep working on it. What definitely is not working is fullfilling its purpose. Underlords subparness ranges from: from not executing commands to doing something completely random. Having no sense for aesthetics and what a proper video should look like or how it should be balanced. In its current state, unusable would be an exaggeration. I have spent over 400 credits for two one minute videos and they are still worse than just having one normal A-Roll video. The feedback is harsh and I want to apologize for that, as the concept and idea is great. Its understanding of video editing and understanding prompts just needs to be greatly improved.