Feature Requests

I know the market is super small but the tool would be amazing, if only I could use it!

We use Descript for all of our user tutorials and our recorded user support sessions. All of these are transcribed and are on our web site, but the problem for a user is finding answers, as people don't have the time to watch a 30 minute video to find the 10 seconds that will solve their problem. We would very much like a LLM style user-facing web interface (like Google's Notebook LM) that we could embed in our web site. This would answer natural language questions from users in the same way that NotebookLM does, with links to the various Descript video files at the point where the LLM is referencing the original file. Descript is uniquely suited for this, as we already have all of the video information in Descript projects. The LLM should only be trained on the Descript projects that I select, with an optional text input so I can add in a block of text that represents our written documentation.

When an input device uses a higher sample rate of 96kHz, rather than the required 48khz, the difference in sample rate can cause an mp4 file to not be produced with audio recorded at that high of a sample rate. Basically - you will not be able to download the combined .mp4 file and will instead need to download cloud recordings and edit/combine manually with audio files. Create a warning/notification for host in Squadcast sessions when a less than ideal sample rate is detected.

Hey the password version does not allow logins now - via email 2FA does work. I changed the password to even test this, but it does not work

instead of having to download your project and re upload to make social clips. Please make existing projects editable from the Workspace.

When wearing headphones, the green screen can fail to apply to the area between your head and the headphone. The result is a very distracting contrast. Uplifting the current green screen functionality (which is great) would be fantastic.

Lots of podcasters and radio producers are using Descript to help with their editing, with the primary focus being on the audio. It would be great if users could load the finished audio into Descript, load in the original video, and Descript then cuts the video to match the audio transcript. MKII of this idea would be to automatically insert cutaways to hide the edits.

You likely already know how great It'd be to have a centralised place to save media. But how great would it be to further squeeze that library's utility via semantic search (from transcripts maybe) so users can access media about relevant topics for videos in whatever terms come to mind?

The title says it all -- there is now no way to affect the automatic room tone on gap clips even when the audio atmosphere changes and likely improves as the recording goes on and there is much more background noise than previously. Also, how does it make sense that when applying Studio Sound, all of the speech is enhanced while all of the spaces in between are not? What, then, is the proposed guidance for the editor? I find Studio Sound un-usable, but this should be considered for the people that do use it.

I need Subtitles with time codes relative to the active scene or selection (0m00s) instead of the absolue time for the whole composition. I use the output for subtitles in video editing not done in Descript. Right now a work around is to move each scene into a seperate composition . But that's a lot of extra work and reduces the benefits of having a single composition for all of the editing.