Clearer wait times and more reliable creation
StudioAPISpeech
- Know how long your generation should take - Studio now shows an estimated time remaining in the player and history. If a generation takes longer than expected, you’ll see delay guidance and a direct way to contact support.
- Keep creating after you close the tab - Start a video with text-to-speech in Studio Lite and leave the rest to sync. labs. Once accepted, speech creation and video generation continue without keeping the tab open. Studio also recovers accepted speech jobs after a reload instead of creating the same audio again.
- Download API results again, without regenerating - Request a download URL for a completed generation using its ID and your API key. No Studio project is required. Retrieve the existing output while it’s available, without paying for another generation. Explore the API
Improvements
- Correct a speaker selection by dragging or resizing its box, or place a box yourself when detection misses the face.
- Studio result previews and signup onboarding load with less unnecessary work.
- Saved video posters now match the current output and plan, so finished results can appear before the video loads.
- Connected ElevenLabs accounts now load the full available voice library, including workspace-shared voices. Restricted voices show why they cannot be used.
- Disabled Sync and speech Preview controls now explain what’s missing and offer retry guidance when eligibility checks fail.
- API keys can create and check transcription jobs and dialogue-edit previews.
- Sign-in codes now appear first in email subjects and previews, making them easier to find without opening the message.
- Plan comparison cards make watermark availability clearer while preserving your plan’s prices and limits.
- The profile menu now works as a named button for keyboard and screen-reader users.
Bug fixes
- Fixed scheduled Studio Lite generations staying stuck after their input uploads finished.
- Fixed repeated Studio submission actions creating duplicate generations.
- Projects with pending or processing generations can no longer be deleted while that work is active.
- Fixed completed assistant uploads being lost after moving to audio input or reloading, and late uploads changing a different session.
- Restored right-click actions on finished assets in desktop list view.
- Fixed restored drafts showing the wrong upload preview or selecting a model that is no longer available.
- Fixed delayed video analysis being applied to a different draft or input.
- Fixed video-only timeline trims failing when the browser had no audio encoder.
- Fixed the space bar losing control of playback after a timeline generation completed.
- Fixed trim-to-limit outputs being rejected because encoding pushed them slightly over the duration limit.
- Fixed voice cloning proceeding with missing samples after an upload failed. Your name and samples now stay in place for retry.
- Fixed cut-off generation estimates and credit reservations using the full audio length instead of the duration being processed.
- Fixed usage totals not refreshing after generation or speech completion, and aligned credit adjustments across activity, history, and CSV exports.
- Fixed top-up dialogs rounding away cents and fallback plan cards showing the wrong voice-clone limit.
- Fixed pasted sign-in codes with spaces or hyphens losing digits.
- Fixed the preview keeping a watermarked result after an eligible plan upgrade, even when a clean output was available.
- Fixed timeline cost estimates changing incorrectly when audio was trimmed or split into clips.
- Fixed voice cloning through API keys failing because the request was not assigned to its owner.