new
improved
fixed
Release roundup—April 16th, 2026
Announcing—the Descript Customer Obsession Hackathon Telethon (May 14–15)
We’re hosting our first-ever Customer Obsessed Telethon—a 2‑day, live, telethon‑style hackathon where we’ll be building alongside real customer feedback in real time.
Have a “biggest little thing” you want added or improved?
Drop it (or upvote an existing idea) on our official Telethon board. You have until April 30th to submit your ideas.
We’re especially looking for:
- Small problems with big impacts: “I should be able to edit Comments” / “Allow me to save custom effects presets”
- Issues that slow you down day-to-day
The hackathon-telethon begins May 14th.
New Transitions
We’ve refreshed the Transitions experience to make it faster to find the right look between scenes. Transitions now come with animated previews, and are organized into a new grid layout with thumbnail previews so you can see the effect before you apply it.
More transition options:
We expanded the library to 16 total transitions, adding:- Clock Wipe
- Shape Wipe
- Double Linear Wipe
- Move In & Out
- Push-Pull
- Scale
- Flip
- Split
- Shake
We added a bulk apply option if you want to use the same transition across all scenes at once to quickly add consistent polish. Many of the new transitions come with additional customization controls so be sure to check the advanced settings.
Just click the transition icon between scenes (or use the Properties panel) to open the new transitions panel.
Live captions in Rooms

Live (real-time) captions are now available in Descript Rooms. When enabled, Rooms will display a real-time caption overlay so you can read along with what’s being said during your session.
You can find it on the Rooms toolbar next to the
Comment
, and Teleprompter buttons.
Captions will appear as an on-screen overlay in your Room, which you can move around as needed. This update improves accessibility for participants who prefer or need captions, makes it easier to follow along in noisy environments or with quieter audio, and helps everyone keep up with fast-paced conversations and reduces mishearing.Also new in Rooms
- Background noise reduction now defaults to off for new users. Hosts and producers can toggle it for all participants from host controls in the sidebar
- If a camera drops and reconnects mid-recording, a recording restarts automatically.
Drive media in the media sidebar
You can now browse and import files from your Media library directly from the Media panel in the sidebar without leaving your project. If you’re on a Business or Enterprise plan you can also access your Brand Studio assets from this panel.
How to find it
- Open the Media panel from the sidebar
- Select My media to access you media library
- Select your files or drag them directly into the Composition window
New in Gen Video—Seedance 2.0

Seedance 2.0 is now available as a video generation model for all users. Seedance 2.0 has been noted for character consistency, motion physics, and multi-shot generation.
How to find it
- Open the AI Tools panel → Generate media
- Select Video, and then Seedance 2.0 from the model dropdown menu
Descript API in open beta
The Descript API beta is now available to all users. You can generate an API token from your account and use it to connect Descript to your own tools and workflows. It allows you to import media, create and edit projects with Underlord, and more all from a terminal or GPT of your choice.
Getting started
- Click here to get your API tokens.
- Read our help center for full instructions
The API also supports Model Context Protocol (MCP) connections so you can take actions in Descript using prompting from a GPT like Claude.
Notes
- As this is a beta, some endpoints and behavior may change over time.
- For advanced documentation, click here.
- If find that trying out the API results in rapidly burning through your AI credits, reach out to support so we can help. We want to encourage experimentation while this feature is still being tested.
Other updates
- We’ve made some backend improvements that should result in better performance for all users in the Editor, making UI interactions more responsive and rendering the timeline faster.
- Underlord now uses reasoning models for complex tasks like script writing, episode planning, and multi-step edits; giving it more room to think through a problem before responding.
- Uploads now automatically resume from where they left off after a network drop, laptop sleep, or network switch. Large files no longer need to restart from scratch.
- Settings → Invoices now shows the upcoming invoice plus the last 6 months of paid invoices by default, with a "load more" button to see older history.
- Locked layers can now be unlocked directly from the canvas no need to open the Layers panel.
- The Hedra Character 2 avatar video generation model has been turned off for all users. Hedra is discontinuing the model, and Kling is the new default.
Bug Fixes
Editor
- Fixed a bug causing CMD + Arrow key layer repositioning in the Scene Editor to revert after clicking away
- Fixed a bug causing layers to get stuck to a scene boundary, preventing trimming or repositioning
- Fixed a bug causing clips to jump to a different scene when edited with the Blade or Trim tool
- Fixed a bug causing a crash when creating a new composition from a text selection in certain projects
- Fixed a bug causing a crash when opening certain layout pack projects
- Fixed a bug causing drag-and-drop to stop working for project files when thumbnail view was enabled in the project panel
- Fixed an issue causing playback to snap back to the same timestamp in certain compositions
- Fixed a bug causing the multicam camera selection on a scene to change unexpectedly when an adjacent scene was deleted
- Fixed a bug preventing right-click scene deletion when the following scene was created from a template
- Fixed a bug causing the Remove from Transcript keyboard shortcut to delete audio instead of only removing transcript text
- Fixed a bug causing the Restore Removed Media keyboard shortcut to also trigger the ignore action simultaneously
- Fixed a bug causing Correct All to apply a previous replacement text when used in a second correction without closing the dialog first
- Fixed a bug preventing transcript corrections from removing just a space between two words
- Fixed several bugs causing transcript corrections to fail with "There was an error changing the text," including words adjacent to an Overdub, corrupted alignment data, and certain projects with special characters
- Fixed a bug causing the app to crash when making bulk transcript corrections across many instances
- Fixed a bug causing audio clips copied between compositions to become uneditable — they appeared dark in the timeline and threw errors on any edits
- Fixed a bug causing the Remove Ignored Text action to reset scene layouts and unexpectedly remove scenes
- Fixed a bug causing the wrong paragraph to be ignored when Wordless Media was enabled
- Fixed a bug preventing an ignore action from being undone when extra whitespace was part of the selection
- Fixed a bug causing the sequence popup not to appear when importing multiple files with the same duration
- Fixed a bug causing spacebar to play the main composition instead of the open sequence when navigating into a sequence from the project files panel
- Fixed a bug causing double-clicking a volume keyframe on a sequence track to open the sequence instead of resetting the keyframe to 0 dB
- Fixed a bug causing the Sequence Editor to open at an incorrect timestamp on the second navigation into a sequence
Speaker detection
- Fixed a bug causing all downstream speaker labels to be cleared when dragging a speaker label across a word
- Fixed a bug preventing speaker labels from saving during the Identify Speakers workflow, causing them to revert to "Speaker #"
- Fixed a bug causing speaker detection to spin indefinitely without completing
- Fixed a bug causing speaker labels to revert after adjusting Studio Sound intensity
Remove Filler Words
- Fixed a bug causing Remove Filler Words to only detect "um" and "uh" when run from the AI Tools panel, ignoring other selected filler words
- Fixed a bug causing filler words to appear grayed out and unselectable in compositions that contained a sequence
- Fixed a bug causing the playhead to jump to a random position when navigating quickly through filler word results
- Fixed a bug causing Remove Filler Words to fail with an error in projects with Wordless Media enabled
Rooms
- Fixed a bug causing push-to-talk to show "Connecting (partial)" and fail for producers
- Fixed an issue causing a guest recording to display an incorrect duration
Screen recorder
- Fixed a bug causing the screen recording track to be truncated after a crash recovery, with the final portion missing
Export
- Fixed a bug causing timeline exports to fail with "Cannot export: cannot read duration"
- Fixed a bug causing exports to fail immediately with an I/O error for certain AIF files
- Fixed a bug causing the Poppins font to render incorrectly in exported video
- Fixed a bug causing video files exported for DaVinci Resolve to show as Media Offline and fail to relink
- Fixed a bug preventing publishing to Web Link, YouTube, and Buzzsprout in certain projects
Drive
- Fixed a bug causing large file uploads to fail partway through
- Fixed a bug causing stock media search results to be capped at 25 regardless of plan level
AI Tools
- Fixed a bug causing AI-generated speech to occasionally come out reversed
- Fixed a bug preventing avatar assignment to a speaker in certain projects
- Updated Draft Show Notes to generate more concise chapter titles
Billing
- Fixed an issue causing the Settings → Usage tab to show missing AI credit history for certain date ranges