Edify Non-linear editing system

1. AI Speech Recognition Feature: AI Speech Recognition Location: Right-click a video clip that contains audio, then select “Audio recognition.” How it works: The user selects a Whisper model and language, starts recognition, and Edify generates a timed transcript from the clip audio. The result can be used for subtitles and text-based editing. Generated content: Speech-to-text transcription. The feature only processes media selected by the user. 2. AI Intelligent Keying Feature: AI Intelligent Keying Location: Right-click a video clip, then select “Keying.” How it works: The user enables AI keying and adjusts the keying region, threshold, edge softness, edge bias, and opacity. Edify uses an RVM video-matting model to generate an alpha matte and remove the background. Generated content: Per-frame foreground alpha mattes. Processing is initiated by the user. 3. AI Beautification Feature: AI Beautification Location: Right-click a video clip, then select “AI Beauty.” How it works: The user adjusts skin smoothing, skin whitening, face slimming, body slimming, and eye enlargement. Edify uses local face-detection and landmark models to track and process the selected person. Generated content: Modified video frames with AI beauty effects. The user must explicitly enable and adjust the effect. 4. AI Dubbing Feature: AI Dubbing Location: Right-click an audio track or commentary clip, then select “Dubbing.” How it works: The user enters text, selects a voice and TTS engine, generates speech, and inserts the generated audio into the timeline. The user can preview and remove the generated audio before export. Generated content: Synthetic voice audio generated from user-provided text. 5. Intelligent Live Commentary Feature: Intelligent Live Commentary Location: Right-click a video clip that contains audio, then select “Live Commentary.” How it works: Edify uses speech recognition to create timed sentences and commentary marks. The user can edit sentences and timing, then apply them to the timeline, generate subtitles, or use them for dubbing. Generated content: Timed commentary text and sentence-level timeline marks. All five features require explicit user action. Edify does not automatically generate AI content in the background. Users select the source media, enable the feature, review the result, and can remove or modify generated content before exporting.