By clicking “Allow All Cookies”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts. View our Privacy Policy for more information.
Minor artifacts with Local AI: When using Local AI to replace audio, there may be very subtle artifacts. These are not present when using Cloud AI.
Background ambience mismatch: If you're replacing a long audio clip with a shorter one, the surrounding background sound may not blend perfectly, leaving an unnatural quiet patch.
Noticeable transitions: The change between original and replaced audio might sometimes sound a bit abrupt or too obvious.
🎥 AI Synthetic Webcam & Lip Sync
2-minute limit during beta: Synthetic Webcam is currently limited to projects under 2 minutes. This is a temporary beta limitation.
Occasional lip sync glitches: You might notice small visual glitches like a doubled chin. It’s rare, but it can happen with both Local and Cloud AI.
Slight quality drop on Local AI: Synthetic Webcam quality using Local AI might not be as sharp as the Cloud version.
Slower performance on some Intel laptops: On Core Ultra 2 laptops, Local AI generation runs slower (about 5x real-time), compared to Arc GPUs (1 to 1.5x real-time).
Intel GPU selection issue: If your Intel system has both integrated and discrete graphics, the AI might default to the integrated one. As a workaround, try disabling the internal GPU in Device Manager.
⚙️ General Local AI
Small cloud request for language detection: Even though Local AI is designed to work offline, a short server request is made to detect the language of your recording.
🌍 Language Support
English only (for now): AI Audio Retakes and Synthetic Webcam currently only work with English-language recordings. Support for other languages is coming soon.
