More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Voiceover Sync: Screen Recording to Onboarding Video

    1

    Syncs a voiceover to a silent screen recording so every click lands on its spoken word, with captions, chapters and automatic QA.

    $9

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Voiceover Sync: Screen Recording to Onboarding Video

    Voiceover Sync: Screen Recording to Onboarding Video

    Example session with this skill installed

    Here is my ElevenLabs voiceover (voice.mp3) and a screen recording of creating a workspace, project and workflow in our app (demo.mp4). Sync them into one smooth onboarding video.

    • Read your context and instructions
    • Compiled the voiceover sync

    final.mp4 (101 s, 1920×1054, H.264 High/yuv420p) with burned-in captions, 6 chapters and −16 LUFS audio; captions.srt; a QA report: "codec OK, drift +0.00 s, freezes > 1.5 s: 0, speed 0.55×–1.71×, loudness −16.4 LUFS, ISSUES: none"; and a cue sheet with one frame per spoken click word, 10/10 matching.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    You recorded your product demo. You generated a voiceover (ElevenLabs, any TTS, or your own voice). Now they don't line up: the dialog opens 30 seconds before the narration mentions it, the screen sits frozen while the voice explains concepts, and the clicks happen at random moments.

    This skill fixes that. The audio stays untouched. The video is retimed, cut and crossfaded around it so that every on-screen action happens on the word that describes it.

    What your agent does

    1. Transcribes the voiceover with word-level timestamps and writes clean captions.
    2. Maps the recording: scene changes, 1 fps contact sheets, 0.2 s close-ups to pin each click, and a crop box if the recording shows an editor around the real content.
    3. Builds a JSON timeline that anchors each click to its cue word ("click Create Workspace").
    4. Renders with ffmpeg: speed-matched clips, crossfaded jump cuts, slow push-in stills for narration with no matching footage, burned-in captions, chapters, and loudness at −16 LUFS.
    5. Scores the result (codec, length drift, freezes, speed range, loudness) and produces a contact sheet with one frame per cue word, then loops until every cue matches.

    Rules built in from real feedback

    • No frame is held longer than ~2 s, because viewers call that "the video froze".
    • Full-window shots for overview narration, because zoomed shots read as "not the whole screen".
    • Speed stays between 0.5× and 2×: editor zoom-out bounces are cut instead of fast-forwarded.
    • Output is H.264 High / yuv420p, which plays in Windows Media Player, QuickTime, browsers and phones.
    • Captions are burned into the picture, because desktop players ignore soft subtitle tracks. A soft track and an .srt for YouTube are included too.

    Measured on a real onboarding video: 151 s recording + 100.5 s ElevenLabs voiceover → 101 s video, 10/10 clicks within 0.4 s of their cue word, no freezes over 1.5 s, speeds 0.55×–1.71×, −16.4 LUFS.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Workflow Automation.

    Needs a silent screen recording and a finished voiceover; it does not record the screen or generate the voice. The agent picks the click anchors by looking at contact sheets, so it is guided rather than fully automatic and takes a few render loops. It cannot fix content mismatches, for example the voice naming something different from what is typed on screen. Footage that is itself low resolution stays soft.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 6 days ago

    Software
    FFmpegffprobe (with libass)Python 3.10+faster-whisper

    On Windows install FFmpeg with: winget install Gyan.FFmpeg. faster-whisper downloads a ~500 MB speech model on first run.

    Permissions

    Allowed Hosts

    huggingface.co (one-time faster-whisper model download)

    File Scopes

    the
    voiceover
    and
    recording
    you
    provide
    a
    work/
    folder
    for
    intermediate
    files
    output
    video

    Everything runs locally; your audio and video are never uploaded.

    Listed6 days ago

    What's inside

    Frequently Asked Questions

    No. The audio is fixed and only normalized in loudness (linear, no pumping). All timing changes happen in the video.

    Yes. Set "burn_captions": false in the config. The soft caption track and the .srt are still produced.

    No. The scan step finds the inner preview, and the render crops to it.

    The scripts are MIT-licensed on GitHub. This listing adds a reviewed package, support, and updates.

    New in Workflow Automation