Nanosaur2-670M
well9472 · Kit
Makes illustrations from text prompts with a small 670M-parameter diffusion model, run through ComfyUI. Built as a low-budget research model for people who wan…
Turns a topic into a narrated motion-design film where a cartoon presenter talks in time with your voice. A Claude Code skill for creators and brands, with your logo on every film.
Talking Explainer is an open-source project on GitHub in the Marketing & SEO category, listed by Mindket. It is free. This is version 1, last updated on . Install it from GitHub as its README says. Get Mindket’s setup notes on your Mindket account, as a zip.
winget install OpenJS.NodeJS.LTS Python.Python.3.12 Gyan.FFmpeg py -3.12 -m pip install numpy pillow scipy faster-whisper
Talking Explainer is a skill for Claude Code. Give it a topic and it produces a narrated motion-design film with a cartoon presenter who talks. The mouth matches your recording, the pose changes on words you pick and the eyes blink. On-screen text and scenes land on each spoken word, and your brand appears at the start, stays in a corner and closes the film.
It suits creators and brands that want explainer films in their own voice and look. A film takes a long session of one to three hours, so plan for that.
| Industry | Who uses it | How they use it | Example |
|---|---|---|---|
| Marketing | Brand teams and solo marketers | Explain a topic in their own voice with their logo on every film | An explainer on how email marketing works |
| Education | Teachers and course creators | Turn a topic into a narrated lesson with words timed to speech | A short lesson on one concept |
| Content creation | YouTubers and short-form creators | Ask for a vertical 9:16 version for TikTok, Reels and Shorts | A vertical version of a finished film |
You can skip it if you will not record your voice or clone it, since each film needs one of the two.
| Use case | Who | What they get back |
|---|---|---|
| Narrated explainer | Creators | An MP4 at 1920x1080 and 60fps with a talking presenter |
| Vertical cut | Short-form publishers | A 9:16 version when you ask for one |
| Branded series | Brand teams | Films that open with your brand, carry it in a corner and end on your call to action |
You need a few things in place first.
On Windows, use PowerShell:
winget install OpenJS.NodeJS.LTS Python.Python.3.12 Gyan.FFmpeg py -3.12 -m pip install numpy pillow scipy faster-whisper
On macOS, use Terminal with Homebrew:
brew install node python@3.12 ffmpeg python3.12 -m pip install numpy pillow scipy faster-whisper
Then add the helper skills on both systems:
npx hyperframes skills update general-video hyperframes-animation media-use
Run the installer from the folder that holds install.mjs:
node install.mjs
Open Claude Code in any folder and ask for a film on a topic. The first time, Claude asks about your brand, your voice route, your speaking style, your opener and closer lines, your presenter and your music. It saves the answers in a profile folder, so later films skip those questions.
make a talking explainer about how email marketing works
For every film after that, Claude asks for four things.
| Tool | What it does |
|---|---|
| talking-explainer | The skill you talk to, which runs the whole film |
| explainer-kit | Its toolbox: checks your setup, keeps your profile, transcribes, handles audio, lip sync and the final sync check |
| spatial-film | The film engine that handles scenes, the camera, the presenter overlay and the render |
Say you want a talking explainer about email marketing. Claude writes out the script as plain text, which you can read aloud or feed to ElevenLabs, then waits for your MP3. You approve the storyboard, and the film is built. You get the MP4 and a credit line for the video description.
If you are new, start with the sample request above, a film about how email marketing works. Claude walks you through the one-time setup, and every later film reuses it.
If you know the field, the timing is the interesting part. Your take is transcribed into word timings. Entrances, camera pushes and highlights are tied to the word that triggers each one, and the mouth shape is picked frame by frame from six shapes per pose. After the render, the film is transcribed again and compared with the script timings.
# Talking Explainer Turns a topic into a narrated motion-design film where a cartoon presenter talks in time with your voice. A Claude Code skill for creators and brands, with your logo on every film. Made by brewed-ops on GitHub. These notes are Mindket's, written from its README; the project's own files and license are in its repository. ## About What it is: Talking Explainer is a skill for Claude Code. Give it a topic and it produces a narrated motion-design film with a cartoon presenter who talks. The mouth matches your recording, the pose changes on words you pick and the eyes blink. On-screen text and scenes land on each spoken word, and your brand appears at the start, stays in a corner and closes the film. Who it suits: It suits creators and brands that want explainer films in their own voice and look. A film takes a long session of one to three hours, so plan for that. ```table Industry | Who uses it | How they use it | Example Marketing | Brand teams and solo marketers | Explain a topic in their own voice with their logo on every film | An explainer on how email marketing works Education | Teachers and course creators | Turn a topic into a narrated lesson with words timed to speech | A short lesson on one concept Content creation | YouTubers and short-form creators | Ask for a vertical 9:16 version for TikTok, Reels and Shorts | A vertical version of a finished film ``` You can skip it if you will not record your voice or clone it, since each film needs one of the two. Where to use it: - Claude Code: the skill and its two helper skills run here - Windows 10/11 and macOS: the computers it supports - ElevenLabs: the route for a cloned voice - TikTok, Reels and Shorts: the places the optional 9:16 cut is made for ```table Use case | Who | What they get back Narrated explainer | Creators | An MP4 at 1920x1080 and 60fps with a talking presenter Vertical cut | Short-form publishers | A 9:16 version when you ask for one Branded series | Brand teams | Films that open with your brand, carry it in a corner and end on your call to action ``` How to set it up: You need a few things in place first. - Claude Code on any paid Claude plan - A voice: an ElevenLabs voice clone on the Starter plan or higher, or a decent microphone in a quiet room - Windows 10/11 or macOS with about 5 GB of free disk, and a dedicated GPU if you want a faster final render - Node.js 20 or newer, Python 3.10 or newer (3.12 is recommended) and ffmpeg 1. Install the tools you lack. Windows uses winget and pip, and macOS uses brew and pip. 2. Add the video engine's helper skills. 3. Get the skill by downloading the zip from the project's Releases page, or by cloning the repository. 4. Start a terminal inside the folder with install.mjs, then run the installer. 5. Restart Claude Code once every line of the machine check says OK. On Windows, use PowerShell: ```powershell winget install OpenJS.NodeJS.LTS Python.Python.3.12 Gyan.FFmpeg py -3.12 -m pip install numpy pillow scipy faster-whisper ``` On macOS, use Terminal with Homebrew: ```bash brew install node python@3.12 ffmpeg python3.12 -m pip install numpy pillow scipy faster-whisper ``` Then add the helper skills on both systems: ```bash npx hyperframes skills update general-video hyperframes-animation media-use ``` Run the installer from the folder that holds install.mjs: ```bash node install.mjs ``` How to use it: Open Claude Code in any folder and ask for a film on a topic. The first time, Claude asks about your brand, your voice route, your speaking style, your opener and closer lines, your presenter and your music. It saves the answers in a profile folder, so later films skip those questions. ```text make a talking explainer about how email marketing works ``` For every film after that, Claude asks for four things. - The topic - One MP3 of your recording, sent after Claude shows you the script - A yes on the storyboard, which lists one scene per paragraph - A check of what appears on screen, needed when your own app is in the film ```table Tool | What it does talking-explainer | The skill you talk to, which runs the whole film explainer-kit | Its toolbox: checks your setup, keeps your profile, transcribes, handles audio, lip sync and the final sync check spatial-film | The film engine that handles scenes, the camera, the presenter overlay and the render ``` Example: Say you want a talking explainer about email marketing. Claude writes out the script as plain text, which you can read aloud or feed to ElevenLabs, then waits for your MP3. You approve the storyboard, and the film is built. You get the MP4 and a credit line for the video description. For newcomers and experts: If you are new, start with the sample request above, a film about how email marketing works. Claude walks you through the one-time setup, and every later film reuses it. If you know the field, the timing is the interesting part. Your take is transcribed into word timings. Entrances, camera pushes and highlights are tied to the word that triggers each one, and the mouth shape is picked frame by frame from six shapes per pose. After the render, the film is transcribed again and compared with the script timings. ```flow Research and script in your voice Your recording as one MP3 Transcription to word timings Storyboard approval Scenes and presenter render Sync check on the finished film ``` Good to know: - Several agents work in parallel on each film, so a session lasts 1 to 3 hours - Claude asks before it installs any system tool - The output is 1920x1080 at 60fps - Music is optional: a track you own, a catalog track after a one-time login, or none - The installer has a --dest option for another folder and a --force option that replaces same-named folders it did not create - Created by Kenneth Villar ## On GitHub, 7 November 2026 - Stars: 67 - Forks: 20 - Watching: 1 - Open issues and pull requests: 0 - Language: Python - License: see the repository - Topics: claude-code, claude-skills, explainer-video, hyperframes, lip-sync, motion-design ## Links - Code: https://github.com/brewed-ops/talking-explainer - Releases: https://github.com/brewed-ops/talking-explainer/releases - Issues: https://github.com/brewed-ops/talking-explainer/issues - Star history: https://www.star-history.com/#brewed-ops/talking-explainer&Date ## Get the code ``` git clone https://github.com/brewed-ops/talking-explainer.git ``` Then follow the setup in its README. Nothing in this kit runs by itself.
README.md 6 KB
Listed GitHub projects in the same category that share the most topics with this one.
| Product | Stars | Activity | Language | License | Last change |
|---|---|---|---|---|---|
| Talking Explainer This page | 67 | Active | Python | — | |
| claude-animation-skillsShares hyperframes · claude-skills · claude-code | 417 | Active | JavaScript | — | |
| YouTube AI Motion GraphicsShares explainer-video · claude-skills · claude-code | 159 | Active | HTML | MIT | |
| bighead-collageShares claude-skills · claude-code | 64 | Active | Python | — |
well9472 · Kit
Makes illustrations from text prompts with a small 670M-parameter diffusion model, run through ComfyUI. Built as a low-budget research model for people who wan…
Waqf-AI · Kit
Reads handwritten Arabic text from images and returns it as text. A fine-tuned PaddleOCR-VL model for people digitizing Arabic handwriting.
Wan-AI · Kit
Makes video clips from a text prompt or an image plus text, at up to 720P and 24 frames per second, in the Diffusers format for a single consumer GPU.
andyhuo520 · Skill
Turns one knowledge point into a short narrated film with a source behind every line, from topic to publish pack, inside Claude Code. For teachers, channel own…
artemnovitckii · Skill
Renders short reel animations on your own computer from a one-line prompt, through six Claude Code skills, for creators who have no footage.
sirioberati · Skill
Gives you 74 animated explainer boards in HTML for YouTube videos, plus every prompt and three Claude skills that sync the boards to your voice and export an M…