API Changelog

This page documents notable changes to the AIFrame API. We recommend checking this page periodically for updates.

Added eye_contact_correction parameter: New boolean parameter for the /api/process-video endpoint to enable or disable automatic eye contact correction.
Improved Lip-Sync Models: Deployed updated lip-sync models (e.g., lipsync2 selectable via lip_sync_correction_model) offering higher accuracy and more natural results.
Enhanced Error Reporting: Error responses now include a more structured error field and a request_id for easier troubleshooting. See Error Handling for details.
Optimized Processing Speed: Average video processing time reduced by approximately 40% for most common video formats and lengths due to backend optimizations.
API Documentation Overhaul: Launched new developer documentation portal with Swagger/OpenAPI reference and detailed guides.

Added Voice Selection Options: Introduced voice_generator and voice_id parameters to the /api/process-video endpoint, allowing selection of different speech synthesis engines and specific voices.
Progress Reporting in Status Endpoint: The /status/{video_id} endpoint now includes a progress field (0.0 to 100.0) indicating the percentage of processing completion.
Improved Background Blur Quality: Enhanced the algorithm for the blur_background feature, resulting in more natural and aesthetically pleasing blurs.

Initial API Release: First public version of the AIFrame API.
Core Endpoints:
- POST /api/process-video: Submit videos for processing.
- GET /status/{video_id}: Check job status.
Supported Features:
- Basic lip-sync correction.
- Background blur (blur_background parameter).
- Asynchronous processing model with polling.
- API Key authentication (via X-API-Key header or api_key query parameter).