August 26th, 2026
3 days ago by Kili Technology
Changelog (26.2.2)
New
- Audio is here! Kili now supports Audio as a new data modality, alongside Text, Image, Video, PDF, and Geospatial — currently available in public beta. The Audio interface is purpose-built for speaker diarization and transcription: segment a recording over time, tag each segment with a speaker, and transcribe the content — then review and export the result as structured, model-ready data. Import audio via the app, the SDK, or a cloud storage integration (MP3, MP4, FLAC, WAV supported).
Contact [email protected] to request access or for more information. See the Audio labeling guide for details. - Added Cloud Storage integration support (AWS S3, Azure Blob Storage, GCP Cloud Storage) for Video projects.
Change
- Added a CSV export option for the data shown in the organization Usage tab, scoped to the currently selected metric and time period.
- Added a "Total time" metric in the Usage tab, combining labeling and review time into a single figure.
Fix
- Fixed the "Change class" menu opening in the wrong position instead of anchoring to the click location.
- Fixed a crash when opening a label whose referenced interface had changed; a clear error message is now shown instead.
- Fixed long asset external IDs forcing the layers panel to expand — names are now truncated with the full name available on hover.