We’ve updated our color Color adjustment tools to make it easier to correct lighting and color, and to apply a consistent look across your footage.

The redesigned Color adjustments settings are now in the Properties panel instead of the Effects menu.

We've added Filter presets for quick adjustments:

This update also adds:

We've updated the underlying color model that powers Temperature and Tint. The previous Temperature slider used an incorrect red/blue range; the new model uses the correct yellow/blue range and is properly calibrated. If your project has Color Adjustments applied, you'll see a prompt to upgrade before you can edit those settings. Upgrading is reversible— you can undo it and return to the old behavior if you prefer to leave your settings as-is.

You can now generate media, add layers, and add layouts directly from the Scene editor panel when your current scene is blank - the new blank slate for any scene includes helpful a lightweight overlay with three quick actions:

Media generation now opens in one unified panel for both and generation. You can switch between and tabs in the same place, with separate prompt and history for each media type.

You will also see commonly used settings more prominently (for example, the model and aspect ratio), so you can adjust them without digging through advanced options.

Underlord now does a better job of linking from chat back into your project, so it’s easier to review what it changed or jump to what it found.

When Underlord references something in your project, it can now link you directly to specific places like:

This helps when you ask Underlord to make edits and want to quickly confirm the result without searching through your timeline.

We have made more upgrades to the lip sync. This update is designed to improve how closely mouth movements match the translated audio, and to reduce visual artifacts in the lip-synced video.

Descript has also rolled out a new AI model for Regenerate that better preserves natural intonation and pacing. When you use Regenerate to re-record a word or phrase, the replacement audio should now sound closer to your original speaking style.

You can now export

Scenes

directly. This gives you a straightforward way to save or share scenes outside of your project when you need to reuse them elsewhere, hand something off to someone else, or keep a copy.

When you generate images while creating a

custom avatar

, Descript will now automatically save those generated images to your . You can find them in Project panel → AI Assets folder → Generated Avatar Images

We made a set of contrast adjustments in dark mode to make text and controls easier to read, especially in lower-light environments.

If anything in dark mode is still hard to read, please let us know what screen you’re on and what you expected to see. We’ve also improved our highlight colors which were not well optimized for Dark mode.

When you upload audio or video to your Drive Media Library, Descript will now transcribe the file automatically.

AI model selectors are now grouped in a dedicated menu section, so you can choose and manage things like transcription, image generation, video generation, and other AI models in one consistent place instead of hunting through General Settings.

We’ve added a lightweight model picker to Home so you can quickly choose which AI model you want to use before starting a new creation flow. This makes it faster to switch models without digging through Settings, and keeps your choice consistent across your next actions from Home.

We spent four weeks focused entirely on making recording in Descript more reliable. The short version: recovery success rates improved dramatically, and several crash-causing bugs now have zero reported incidents.

Here's what changed:

Descript has a new look. We used the refresh to audit and improve contrast across the interface—the previous palette wasn't fully WCAG 2.0 compliant in a number of places, and the new one is. That means better legibility for everyone, including users with color vision deficiencies. If you're seeing anything that still looks off, flag it—there are a few known issues still being wrapped up.

Also, probably not important for you but maybe: things are a bit redder now and our brand colors have been updated across the board.

The wait is over, Descript has added the Media Library. You can now upload and organize media at the drive level, making it easier to reuse files across multiple projects, saving you media minutes, as well as time waiting for transcription, and file uploads.

Click Media Library in the left sidebar of your home page. Any media you upload here is shared with everyone on your drive. When you add media from the library to a project, it uses media minutes for transcription. This has been a longstanding feature request, and we’re excited to finally bring it to life.

Import from other projects

You can now preview audio and video before Underlord applies them to your project. When generating media like avatars or video clips, Underlord will show you preview options so you can confirm the results match what you're looking for before committing to the generation.

This helps you avoid wasting time and credits on results that don't meet your needs, especially for time-consuming or resource-intensive generations.

We've improved how layouts work in your projects. When you apply a layout, the editor now better understands which layers should be replaced and which should stay.

These changes make layouts behave more consistently and give you better control over your scenes.

Two new models are now available in Underlord's model picker.

sits between Sonnet 4.5 and Opus in terms of quality and cost. It's Underlord's most precise instruction-follower, making it a strong choice for complex, multi-step editing workflows, creative work built from existing clips, and script-based generation. It had the fewest errors across all test cases. Note that it's the slowest model in the lineup, and for audio-heavy tasks like Studio Sound, Opus remains the better option.

sits between Claude Sonnet 4.5 and Sonnet 4.6 in quality and cost. It's a cost-effective option for longer projects and everyday editing tasks, and leads the lineup on volume leveling. For tasks like removing filler words, pauses, or generating Studio Sound, Claude models will produce cleaner results.

To switch models, open the model picker in Underlord’s settings by clicking the cube icon.

We've added Nano Banana 2 to our image generation model options. This model is built on improved architecture and offers better performance across all aspects of image generation.

You can select this in the Settings menu, or under the cube icon within the generate an image section of the AI tools sidebar.

Rooms now alerts you when your CPU is overloaded during a call. When this happens, you'll see a warning with suggestions to improve performance, like turning off background noise reduction or disabling your video.

Background noise reduction is now easier to access with a toggle in the microphone menu. If your CPU gets overloaded, you'll see a warning, and the feature will automatically disable itself to prevent audio issues instead of crashing.

It has been awhile. Thank you so much for all of the feedback you gave in response to Laura’s letter. You can expect to see ideas taken directly from user suggestions throughout the year - including quite a few below! (Resizable sidebar, British English, better avatars.) Here’s what’s changed since we last posted.

We changed the Descript brand colors from blue to red. We know it might take a moment to adjust to looking for the red logo when you open Descript, but at least we didn’t do this. Instead, the logo is changing to honor the color of the record button.

You can now resize the sidebar in the editor. Just drag the edge of the sidebar to adjust its width. This makes it easier to read longer chat messages in Underlord and view other sidebar content more comfortably. The sidebar width is preserved while the app is open. When you resize your window, the sidebar maintains its width rather than scaling proportionally.

You now have more control when saving scenes as layouts. When you click "Save as new layout" from the scene sidebar or right-click menu, a new panel lets you choose what to do with each layer:

When you turn a layer into a placeholder:

You can now select Claude Opus 4.6 as your model in Underlord. Based on our internal testing, it’s much better especially when working with visual elements and complex multi-step tasks.

To use Opus 4.6, open the model picker in Underlord's settings.

You can now choose between three models when generating Avatars:

The Kling models are better at everything, just to be honest. They are clearer. Less twitching or eye melting. More expensive, but just do yourself a favor and try it. To select a model, open the model picker in Settings or the Avatar confirmation modal.

When using Kling v2 or Kling v2 Pro, you can now add an attitude prompt to guide your avatar's tone and expression. For example, you might describe the attitude as "friendly and enthusiastic" or "professional and calm."

Keep in mind that these prompts have a subtle effect. Audio remains the primary driver of your avatar's behavior.

You can now assign waveforms to individual speakers so they only animate when that person is talking. This can help you create audiograms where different speakers are repped by different waveforms. A customer asked for this and it was super easy to do. The waveform will only animate when that speaker talks.

When you translate and dub your videos, we now default to matching timing between videos. This means fewer videos where the audio feels weirdly sped up or slowed down. If a direct 1:1 translation is more important, you can choose “Direct translation” instead.

When creating a translation with dubbing enabled, look for the Style dropdown in the translation menu. You can select either:

This gives you much more flexibility to control the direction of your video. If you prefer to keep the lip sync consistent and smooth, use match timing. If the details of your content are more important, opting for direct translation will get you more accurate results.

We've also updated our lip sync model for translated videos. It’s hella good now. The new version delivers better visual quality and more reliable performance. The improvements are automatic when you add lip sync to translated videos. You don't need to change your workflow—just continue creating translations and adding lip sync as usual.

We've made it easier to undo unwanted edits, understand what the agent is doing, and view media when using Underlord:

These changes make it more obvious how to control the agent's work and recover from mistakes.

We’ve made viewing projects on mobile way better. This is a small update, but we expect it to be the first of many optimizations we make to accessing Descript across your many devices.

We've cleaned up the mobile experience to hide buttons that don't work on mobile, so you'll only see options that are actually available on your device.

Full editing capabilities aren't available yet, but you can now review and share your work wherever you are.

You can now transcribe audio in regional English (UK). This option appears in the language dropdown when you upload media or adjust transcription settings.

The transcription uses United Kingdom spelling and phrasing. For example, "colour" instead of "color." To use it, select English (UK) from the language dropdown in your transcription settings. You can also re-transcribe existing content between US and UK English.

English (UK) won't be automatically detected when using the "Auto" language option—you'll need to select it manually from the dropdown or set it as your default transcription language in User Settings.

You can now add verified Descript skills to your LinkedIn profile. Based on how you use Descript, you may qualify for a certification badge that displays on your profile.

There are three certification levels based on your feature usage and editing activity: Proficient, Advanced, and Expert. As you continue using Descript and exploring more features, your certification level may update automatically.

You can manage your LinkedIn connection anytime in .

Since the beginning of the Beta, we’ve learned a lot from how you’re using Underlord. Your feedback has shown us both what's working and where we need to raise the bar.

Today we’re to rolling out improvements to Underlord that make it faster, less expensive, and more capable across the board.

You’ll notice more thoughtful edits and stronger responses when using:

Here are a few examples of the kinds of queries where Underlord has improved:

This new version of Underlord is much more efficient when using AI-credits, and better explaining its reasoning, and self-correction:

It’s also better at explaining its reasoning and self-correcting its mistakes:

: You can also customize the AI-model that powers Underlord depending on your use case. To do so, open the model picker and choose from the latest models from Anthropic, OpenAI, Google, and others.

If you’ve got an established, fine-tuned workflow with the old version of Underlord, and want to go back, you can temporarily opt out by opening Settings → Labs, and toggling to OFF. This toggle will be available for the next few weeks.

If you toggle it off, you’ll be presented with a quick survey about your experience. Please answer it to help us continue improving!

Join the Descript Changelog for a special year-end message:

And now, on to the updates.

You can now play your script while recording screen or camera layers on top of it. This makes it easier to record b-roll that syncs with your existing Script track.

To use this feature:

You can now use the teleprompter during your Rooms calls. This feature helps you deliver your content smoothly while maintaining eye contact with your camera.

The teleprompter allows you to read your script directly within the Rooms interface, making it easier to stay on track during recordings or live sessions.

Drive admins can now control who can edit the transcription glossary and "do not translate" list in Brand Studio. When enabled, these settings prevent other users from modifying these glossaries, giving admins more control over consistent transcription and translation across the team.

We’ve made the decision to turn off the YouTube import feature in Descript for now. We want to share a bit of context about why.

Over the past several months, YouTube imports have become increasingly unreliable. Success rates dropped significantly, and many videos simply stopped importing altogether.

YouTube regularly updates its systems to prevent automated downloading. These changes directly affect tools like ours, and they can break or limit importing with little notice. We’ve evaluated short-term and long-term options, but at this point we are not able to provide a version of YouTube import that works consistently over the longterm.

We hope to revisit this in the future, especially if conditions change, but we don’t want to offer a feature that fails unpredictably or creates confusion.

You can still work with YouTube content in Descript you’ll just need to download the videos manually first, then import the files directly into your project. We know this adds an extra step, and we’re sorry for the inconvenience.

If you’re downloading your , the most reliable option is YouTube Studio, where you can download the original files you uploaded.

Here’s how:

Go to YouTube Studio and sign in with the account that owns the video.

Click in the left sidebar to view your uploads.

Hover over the video, click the menu (three dots), and select .

Drag the downloaded file into Descript or add it from the Project panel.

Using YouTube Studio is the most stable, YouTube-compliant way to access your own videos and avoids the issues that affected our previous import feature.

When you translate and dub a video you can now apply lip sync to match the speaker's mouth movements to the translated audio. This makes translated videos look more natural and seamless.

To use lip sync:

Lip sync is only available on our updated plans using AI credits.

Lip sync only applies to faces detected in your video's main camera layers—it won't affect screen recordings or b-roll footage.

When you start a new video with Underlord, you can now a multi-step, multi-question Project brief that outlines the plan before any editing begins. Underlord will propose a direction for your video—including stylistic choices—and wait for your approval before taking action. This helps you feel confident about the approach before credits are spent. You can iterate on the style and direction until you're ready to move forward.

To use it, start with a template, type a prompt, or upload files to an empty composition. Underlord will present a plan and style options you can adjust before it begins editing.

You can now select Kling O1 as your video generation model in Descript. This model generates high-resolution, cinematic visuals and can create videos up to 10 seconds long.

To use Kling O1, open the model picker in your video generation settings or in the Generate video panel.

We’ve also added Flux 2 Pro as an image generation model in Descript. This model balances speed and quality, with improved text rendering and capabilities for restyling existing images.

To switch models, go to Settings → General and select Flux 2 Pro from the image generation model dropdown, or select it in the Generate an image panel using the model picker.

The Learn Descript team has been hard at work making new content to help you navigate our latest features. Today’s video covers advanced video generation. You’ll learn about:

If you’re looking for a more basic tutorial on video generation, check out this video first.

You can now select Gemini 3 as your AI model in Underlord. Google's latest model has been impressive in testing for complex tasks and multi-step workflows. The areas it does well:

To switch models, open the model picker in your Underlord settings.

You can now select Nano Banana Pro as your image generation model in Descript. This is Google's state-of-the-art model for high-resolution image generation and precise editing.

Nano Banana Pro offers many improvements of over the original Nano Banana model:

To use Nano Banana Pro, go to Settings → General and select it from the image generation model dropdown or select it in the Generate an image panel using the cube icon.

You can now extend video clips using Underlord when you need more footage. The new Extend Video option generates additional frames from the last moment of your clip, making it useful when a clip is too short or you need more content without reshooting. Or when you just need to get weird with it.

To extend a video:

The extension uses the last keyframe of your clip as the starting point. If you have scenes in your project, the extension will automatically fill the remaining time in the current scene up to your model's maximum duration.

For best results, choose clips with smooth, steady motion like a person walking or a camera panning. Avoid clips with abrupt cuts or mid-action endings.

You can extend generated clips, recorded media, or imported content.

When you translate a video with dubbing, you'll now see a progress indicator in the sidebar that shows which translations are currently processing. You’ll also see indicators above each section as it’s being translated in real time.

The updated loading state makes it easier to track multiple translations at once and gives you a better sense of how long the process will take.

We’ve introduced some improvements to our language dubbing engine. When you translate and dub a video, the translated audio should match the timing of the original audio more closely. The updated translation system better preserves the pacing and rhythm of your original speech when generating translations, so the dubbed version sounds more like your natural delivery in the target language. This improvement applies automatically when you translate and dub your videos—no additional steps required.

You can now create and save custom templates in Descript. When you find a prompt that works well with Underlord, save it as a template so you can reuse it anytime—no more copying and pasting prompts.

To create a template:

Your saved templates will appear in the template gallery under "Created by me."

Custom templates are private and visible only to you. You can't edit or delete them after creation, but these features are planned for future updates.

Click here to try it now.

Descript now supports 39 additional languages for translating captions and dubbing.

Bulgarian, Filipino, Indonesian, Russian, Tamil, and Ukrainian

English (UK), Spanish (Spain), French (Canada), Portuguese (Portugal), and Chinese (Traditional)

Regional variants apply to captions only—the dubbed voiceover will use the standard language without a regional accent.

Afrikaans, Armenian, Assamese, Azerbaijani, Belarusian, Bengali, Bosnian, Welsh, Estonian, Galician, Hausa, Javanese, Georgian, Kazakh, Luxembourgish, Macedonian, Malayalam, Marathi, Nepali, Punjabi, Sindhi, Somali, Serbian, Swahili, Telugu, Thai, and Vietnamese

You'll find these new options in the Translate menu under AI Tools.

Check out the tutorial on Learn Descript for more.

Add a frosted-glass effect to any shape or text element in your project. Glass blur softens the appearance of content underneath the element, making it useful for directing attention or obscuring things you don't want to show in your videos.

To use Glass blur:

Glass blur is a visual effect only and should not be relied on to securely hide sensitive information. For secure redaction, use a solid shape with 100% opacity instead.

When you upgrade to a different subscription plan mid-cycle, you'll now see the prorated cost displayed during checkout. Previously, the full price was shown even though you were only charged the prorated amount.

This fix ensures the checkout flow accurately reflects what you'll pay when upgrading between plans.

The bug fixes section will now include the effected app surface for easier reference. The bugs in this list largely come from direct user reports.

Drive admins on Enterprise plans can now restrict what assets are used by their team by setting permissions in Brand Studio. This helps you enforce brand compliance across all projects without constant manual review.

When you set permissions, team members will only see approved brand assets in their editor. Your brand colors, fonts, logos, and media appear first in color pickers, font menus, and media browsers—making it easy for your team to stay on-brand by default.

To set up permissions, open Brand Studio from your Drive, go to the Permissions tab, and choose brand assets. This feature is available on Enterprise plans.

For editors on Enterprise drives, you’ll now see a tab for brand-approved assets in your Media panel.

We’ve also added Sora 2 to our generative video options in Descript. Sora 2 creates cohesive multi-shot videos from text prompts, with smooth motion, camera continuity, and cinematic realism.

Sora 2 automatically generates audio including music, sound effects, and speech when you include quoted dialogue in your prompt.

Learn more about the best ways to generate and use video in Descript with our friends at Learn Descript.