What this proves
The panel, live preview, timeline selection, progress log, transparent-frame output, and Premiere placement are observable in the recording.
This uncut screen recording shows preset selection, timeline audio export, transcription, deterministic frame rendering, and caption-media placement from the real Caption Plug panel.
Real Caption Plug product footage. The run uses Groq transcription; it is not presented as a local-Whisper benchmark.
Numbers below are transcribed from the panel log visible in this specific recording. They describe one short run, not a universal speed benchmark.
Caption Plug is open inside a Premiere workspace. The editor checks Beast Pop, CapCut Bounce, Karaoke Fill, and Impact Shake in the panel's live renderer.
The selected clip defines the audio range. Caption Plug exports that range instead of requiring a separate media picker.
The visible run sends the audio to the configured Groq provider, receives 13 words, and groups them into caption ranges. Local Whisper and OpenAI are separate available provider choices; this recording does not claim to show them.
The run renders 139 transparent frames at 576 × 1024. Caption Plug reads the sequence dimensions and frame rate rather than assuming a fixed social preset.
The rendered caption media is imported into the project and placed on a new video track. The source run folder remains beside the project for relinking.
The panel, live preview, timeline selection, progress log, transparent-frame output, and Premiere placement are observable in the recording.
The clip does not compare transcription accuracy, total speed, stability, or correction time against another product.
The current public build is newer than the original footage. See the changelog for every material release change.