Questions,answered plainly.

Your media is not uploaded for AI processing. Account access, licensing, updates, support and initial model downloads may still use the network.

Windows 10 or 11 with 8 GB of RAM works. 16 GB and a discrete GPU make it comfortably faster than real time. Without a GPU it still runs, just slower.

Between roughly 1 and 3 GB depending on the size you pick. Smaller models are faster and lighter; larger ones handle accents and noisy audio better.

MP4, MKV, MOV, AVI, WebM for video and WAV, MP3, FLAC, M4A for audio. Subtitles export as SRT, VTT, plain text, or burned into a new video.

Accuracy varies with the selected model, language, accent, overlapping speech and background noise. The transcript editor lets you correct the result when needed.

Both are in progress. The Windows build is currently in beta testing.