We’ve added a new Smart fill tool for text layers when you apply a layout from the gallery. It reads the script track and automatically fills text layers like titles and lists with relevant details from your transcription.

This is a great way to save time with a better-than-blank placeholder, or final text if you’re satisfied with the results.

This setting is toggled on by default for any of our premade layout packs, but you can toggle it on or off for your custom layout packs from the Layout pack menu in Drive view.

There’s a new illustration style for AI generated media when using both the AI video maker (available from the Home screen) and the Quick design action on the sidebar. The new setting is a brightly colored “collage” art style using more abstracted shapes for a more modern and lighthearted look.

You can now adjust the A/V sync of files automatically from the Project panel. This allows you to fix audio/video sync issues by applying a positive value to delay the video, or a negative value to advance the video.

This should save you time if you've been manually adjusting A/V sync issues using the Sequence editor.

We’re now able to process Green screen (background removal) and Blur speaker background instantly, without the need for long processing times. We’ve also removed the Beta label on Blur speaker background. This update is rolling out slowly over the next few days, so if you aren’t seeing these effects process instantly, check back soon.

We’ve added a secondary button to Generate your avatar to the Scene editor toolbar. This is just to make this option more prominent in the case that you haven’t Generated your avatar yet after adding it to a project.

A few weeks ago we had to disable the ability to Hide the recording dock on Windows when using the Screen recorder in order to resolve some bugs. We have reverted this change - and you will now have the recording dock automatically hidden from your final recordings when using Windows.

We’ve added a toggle to the Remove filler words AI action. You can now toggle on the “Avoid harsh cuts” setting, and our updated modal will analyze each removed filler word to ensure that it can be removed cleanly, without creating a harsh or jarring edit boundary that will be distracting for your viewers.

If a filler word cannot be removed cleanly it will be skipped to preserve the flow of your content.

Keep in mind, this is analyzing for harsh audio cuts, so you may still see video cuts that you need to adjust manually. It also adds a small amount of processing time when removing filler words.

Regenerate, our tool that regenerates sections of your script to make corrections without needing to re-record, can now regenerate sections of your video without creating awkward lip-sync issues.

When you make script corrections or want to adjust the tone, Regenerate video will automatically generate AI speech and video that flows naturally. As always, you’ll need to consent from any speakers included in the regenerated section. Regenerate video can take up a few minutes to process. In the background, we’re regenerating the audio first, and then regenerating the video to match.

Keep in mind that this feature is in beta, and still being improved. At present, it can only be applied to your Script track, and cannot be applied to any other layers.

It works best when the speaker is centered in frame with their mouth visible. It also works best with video files under 4K.

You can now use the Quick design action under the section of Underlord to quickly format your single-scene projects with scenes, layouts and B-roll. Just click the Quick design option and select from AI-generated B-roll, or Stock media. We’ll take care of the rest, detecting your layers and applying an appropriate layout.

This currently only works if you haven’t added scenes to your projects yet. We’ll post an update as soon as this can be applied to projects that already have scenes.

We now automatically apply a default layout to your recordings. This will select a layout based on the number of layers if one is not already selected. You can change to a different layout from the scene panel.

Over the last 6 months, we’ve slowly moved all of our users onto an upgraded version of the Desktop app that provides better performance, stability, and eliminates the need to keep local files for editing.

This new version also gives us the ability to make more frequent updates to fix bugs, implement changes to our backend, and more, without forcing you to update the app.

As a result, our version numbering system is no longer particularly relevant for most releases. We’ll still be posting changelog updates regularly—they’ll just be more of a roundup style to catch important changes, feature updates, and bug fixes.

We’ve made a few improvements to the AI video maker, which you can find on the Home screen. Here’s what’s new:

You can try it right now by clicking here.

We’re using a new and improved image generator when you select “Generate an image” from the Underlord panel on the sidebar. You’ll see an overall improvement in the quality of images. This has not yet been ported over to the AI video maker, so keep in mind you’ll be seeing images generating from the older image generation model when using that tool—for the time being.

We no longer support invite links to Drive, instead we’ve improved the experience for email invites. This helps keep Drive invites more organized and easier to track whether or not they have been accepted.

We’re rolling out a fully redesigned, and reconsidered Timeline view in Labs this week. Labs is our testing ground for new and exciting features that are still in development. You can toggle on this new Timeline experience by opening the Settings menu → General → and scrolling down to the Labs section.

The new timeline comes with many changes, which is why we’re putting it in Labs for users to get acquainted, and to allow more time for testing and polish.

Here’s what you need to know:

You can read more about all the changes in our help center, including a video walkthrough from Andrew.

We’ve re-built and relabeled our AI video maker adding a host of new features including Avatars, automatic layouts, captions, titles, scenes and even b-roll.

Select AI video maker from the Home screen to get started. From here you can start generating your video by:

If you select either , or you’ll first select an Avatar to be your presenter, along with an AI speaker to voice it. If you select “Upload a file” - we’ll take you straight to generating b-roll, or auto-selecting stock media for your b-roll.

Once you’ve made your selections, the video will start generating, and you can either Export immediately, or dive in and edit it. If you’re using a pre-recorded video, but prefer an Avatar, you can always swap it out inside the project.

This tool is a great way to bring a quick idea to life, or just try out Descript’s editing features with no barrier to entry in terms of having an idea or pre-recorded material.

Development of our new Screen recorder is moving forward. This week, we’re removing the Beta label.

Even more updates are coming to the screen recorder, so stay tuned.

Back in February, we rolled out major improvements to speaker quality to our Business and Enterprise users. We have improved the quality even more and are deploying both sets of improvements to all users in the coming weeks. You should notice a higher overall audio quality with better, more natural delivery. This also means that we will soon be fully deprecating custom speakers made with our very old models. These are labeled as "Removing soon" in the AI Speakers menu in Drive view. If you’re still using one, you’ll want to create a new custom speaker with the upgraded models by 6/30/2025. You can do this by clicking the "New speaker" button in the top right of the AI speakers menu in Drive view.

We've begun our rollout of avatars. Over the next week, you'll start to see avatars mentioned on our Home screen and in the speaker label. With avatars, there's no need to stress about recording yourself for every single video you have to make. Avatars let you make video that’s just as informative and engaging—in a fraction of the time, with no recording at all.

Just write your script, select a voice from our new stock voices and then choose your avatar—from the variety of options in our gallery, or from an image you upload.

You can also start an Avatar based project from the Home screen, either by selecting Text-to-AI Video and ticking the Avatar box, or by selecting Create AI Avatar. Or just click here.

Custom avatars work best with human subject in center frame so we suggest avoiding images of animals or inanimate objects.

For more details on best practices when creating custom Avatars, our Help center has you covered.

To make Avatars even more robust, we added 13 new stock speakers that will help give your avatar just the right delivery. We've welcomed five new conversational voices, four creator-style voices, another British voice and a few others.

Say hello to the gang.

We’ve brought back the ability to import video directly from YouTube links. This is a very popular feature that we want to support long term.

For users with specific transcription needs, or who run recurring issues with some transcriptions, we’ve added a new Transcription provider selection in the Settings menu under the Advanced section. We do not recommend that most users switch this away from the Automatic setting, but in some cases it may be worthwhile to try a different option.

The options at present are:

We may add further options in the future.

Here's what's new in Rooms:

For users who have been upgraded to the new Desktop experience, we finally have a replacement for the . The new

Screen Recorder

will launch to all users on the upgraded Desktop within the next week, bringing enhanced functionality and performance to make your recording experience smoother and more versatile. This update is just the beginning, and we’re launching it in a Beta state while we make improvements and round out the feature set.

Like our web recorders, it uploads your video in the background while you record for instant playback when you hit stop.

At the moment, it does not include screen area selection, or multitrack file recordings.

Both of these features are in the works and will be added when the Screen recorder leaves Beta.

If you’re on the old Desktop, you can upgrade manually using the . By May 15th, 2025, all users will be on the new version so you might as well jump in now.

We're building a new way to use AI to edit your videos in Descript. If you want to help us break it and build it better, apply to get exclusive early access. Learn more and apply here.

We're excited to introduce a new Home screen in Drive view, making it easier than ever to start your next project. When you open Descript, you'll find a streamlined landing page with quick-start options for all your creation needs - whether you want to edit a video, make a podcast, create social clips, transcribe a file, clean up audio, or remove backgrounds.

The new Home screen is designed to help you find what you need faster, including a Recent projects view. You'll also discover our popular features front and center, including Translation & dubbing capabilities. These will rotate over time as new features are introduced.

The hope with this update is to remove the uncertainty of a blank page project and putting the full power of Descript at your fingertips from the moment you open the app. The new Home screen is rolling out to all users in version 112.

A few weeks back we created a new AI Video maker in the ChatGPT store. This tool automatically generates a video explainer on a topic of your choosing, with customizable style and tone. Now, we’ve brought this experience into the app, available from the new Home screen, so you can create narrated, illustrated video with just a few clicks. It’s a great way to experiment with the tools Descript has to offer, or make a quick piece of content for easy sharing.

We're introducing a redesigned Editor recorder experience in version 112. The recorder now lives in the play dock instead of the sidebar, making it more accessible within your workflow. When recording with an empty script, the recorder takes up your full screen for better focus. If there's content in your script, you can see it while recording - the recorder will only occupy the Scene Editor area. The redesign aims to make recording more intuitive and accessible, integrating seamlessly with your workflow while maintaining visibility of important content during recording sessions.

Stable frame is a new feature that intelligently positions the playhead to show the most representative frame when selecting scenes or layers. When clicking on scenes from the scene rail or script boundary, or layers from the canvas, the system accounts for animations and scene transitions to display the most stable and useful frame (i.e. one that is not in the middle of animating.)

These stable frames will also be used for the Scene thumbnail to make scene rail more useful as a visual representation of the storyboard or through line of your project.

This enhancement addresses a common user pain point where the playhead might land mid-animation, resulting in a confusing or unclear view of the scene.

We’ve added two new placeholder layer types to the Elements panel for Camera and Screen video. These will to help you keep your project organized as you’re assembling layouts, and make it easier to remember where you will eventually place your content.

The three placeholder layer types are now labeled:

You can adjust these layers like any other Scene element, applying animations or transitions. Once you're ready to add your media, you can do so from the Scene-editor → Add media.

Here's what's new in Rooms:

We’ve made some big changes to working with Scenes and adding Layouts. First, you’ll notice that the term “template” has been replaced with the term Layout pack. A Layout pack contains individual layouts which are organized by Scene-type (i.e. Title, Lower Third, 3-camera etc.)

As a refresher, Scenes are divisions in your timeline that represent changes to your visual layout.

We’ll preselect a default layout pack, but you can click Change layout pack to visit the Gallery, view your Private custom layout packs, or view packs available from your Drive. Each layout has a number of different styles that you can view by hovering over the Layout card and clicking See More.

There’s a handy search bar, and we pre-filter your layouts to only show layouts that match the amount of layers in your composition (i.e. If you only have 1 camera file, we won’t show multicam layouts unless you toggle off the filter.)

The 3-dot menu on this panel these options:

If you don’t see them right away, you will by the time version 109 is released.

We’ve added a new Scene toolbar under the Canvas anytime a scene is selected. You can use it to swap layouts on the fly, adjust transitions, or add backgrounds to your scene. This toolbar also shows the current layout-type for easy reference

If you’re having deja vu, this was announced and rolled back a few months ago.

There is also an option to Add scene from the Script toolbar. This will take you straight into the new Layout picker panel so you can quickly build your project from scratch before you’ve added any content or quickly apply layouts to you script by highlighting each scene-to-be.

We’ve refreshed our collection of premade Layout packs (FKA templates,) adding newly designed packs, redesigned packs, and removing some layout packs that simply won’t be missed.

Here are some of the new packs, which sound tasty and look great:

You can locate the gallery by clicking "Change layout pack" from the scene panel.

With all the updates to working with Scenes and Layouts, we’ve added a new type of Transition animation that was created to save you mountains of time. Smart transitions apply automatically to all visual layers in a scene, and select appropriate crossfades, zooms, wipes etc. to the Scene transition. This transition effect comes applied automatically to some of our default layout packs, or you can add it manually from the Scene panel on the sidebar.

You can now create translated dubbing and captions for Chinese (Mandarin,) Japanese, and Korean. These are in the same list along side our existing translation options. These languages also feature native sounding speakers for Business and Enterprise plans.

Eye contact - our tool that draws the speakers eyes to the camera at all times - is now out of Beta. This just means that we’ve observed the number of bugs and other errors to be low enough that the feature is considered stable. As an added benefit, this effect now processes instantly when using the Web or upgraded Desktop version of Descript.

Business an Enterprise users will be getting an improved training model when generating AI speakers. This updated and improved model creates more natural sounding custom AI speakers. Any existing voices will be retrained using the new speaker model so the benefits are back-ported.

For Enterprise users:

On February 11th, our custom AI speakers are getting a major upgrade that will deliver voices with more natural speech with better intonation and pacing*.

These upgrades also bring faster text-to-speech generation, and add support for custom AI speakers in 19 languages: English, Spanish, French, German, Italian, Portuguese, Polish, Croatian, Czech, Danish, Dutch, Finnish, Hungarian, Malay, Norwegian, Romanian, Slovak, Swedish, Turkish.

You can hear a before and after sample below. This voice was trained first on our old model, and then retrained with the upgraded model. Have a listen:

If you prefer to opt out you can let us know by submitting a ticket to our support team.

*Please note that these upgrades will apply to Text to Speech - not Regenerate.

You can now add a blur effect to the background of speaker focused videos. This can be toggled on from the Layer menu for any video track, and will blur out everything behind the speaker. There is an intensity slider for more or less blur. Please note, this only works with human speakers. Pets, inanimate objects, and ghostly apparitions will be blurred as if they are part of the background.

Generating text-to-speech is now supported in 19 languages, and we will auto detect the language of the text that is entered. You can enter TTS in: English, Spanish, French, German, Italian, Portuguese, Polish, Czech, Dutch, Romanian, Turkish, Danish, Malay, Finnish, Hungarian, Norwegian, Slovak, Croatian, Swedish.

Our translation powers when using stock speakers has expanded from 5 supported languages, to 21 supported languages. The full list of languages that support dubbing: English, Spanish, French, German, Italian, Portuguese, Hindi, Polish, Czech, Dutch, Romanian, Turkish, Greek, Danish, Malay, Finnish, Hungarian, Norwegian, Slovak, Croatian, Swedish.

For Business and Enterprise users, the following languages will also have the benefit of our new Recommended voices which are designed specifically for them: Spanish, French, German, Italian, Portuguese (BR), Polish, Dutch, Hindi, Turkish, and Swedish.

This brings the total number of languages with Recommended voices to 10.

With our first update back from the holidays, we’re adding a new way to export your translated compositions in bulk, retiring jibber jabber, making some small tweaks in the editor, and fixing a number of bugs.

You can now export all of your translated compositions in a batch process. We’ve added checkboxes to the Language selection list on the Export tab, so just select all the translated versions you’d like exported, and they’ll be automatically queued up and processed. A great time saver on the path to expanding your audience.

We’ve made it easier to understand when you’re nearing your text-to-speech limit by eliminating Jibber Jabber (replacing random words with nonsense when you’ve used your balance.) Now, we’ll simply stop generating new TTS, and you can increase your monthly limits by upgrading if you need more time. You’ll hit this prompt and click or dismiss it.

You’ll only hit this once, and then be met with a smaller prompt as a reminder that you are out of text-to-speech minutes.