E-learning Video

English listening

English listening practice: reader, vocab, guess and joke formats

Picture shape: 9:16
Runs on Claude CodeCodexAntigravity

Turn one short English lesson for Vietnamese learners into a 15-80 second vertical listening video (1080x1920), in one of four formats, voiced for free on your own computer.

Important: this is a skill for terminal AI coding agents (Claude Code, Codex) that runs on your own computer. It is not a prompt for the ChatGPT website. Check the requirements below before buying.

Four formats

  • Reader: a short A1-B1 story of 6-16 sentences, read line by line on cream paper with a pencil-sketch picture card. A bilingual English and Vietnamese list scrolls, a blue band marks the sentence being read and a box jumps from word to word with the voice. About 1.9 seconds per line plus 3 seconds.
  • Vocab: 1-4 pages of a passage, with 3-6 target words per page tinted by word class and a small Vietnamese gloss under each, a waveform pill and a picture in a glowing frame. About 8 seconds per page.
  • Guess: 5-8 words of one topic: the meaning, the first letter and blanks, an ink countdown, then the answer card with IPA. About 7 seconds per word plus 4 seconds.
  • Joke: one English pun in 2-6 chalk lines on a slate board, a Vietnamese caption per line and a hand-drawn dog mascot that snickers at the punch line. About 14-18 seconds.

How it works

Open your agent and type a request like this:

Làm video luyện nghe tiếng Anh dạng reader, trình độ A2, câu chuyện ngắn về một buổi sáng đi chợ
  1. The agent writes one lesson into data/lesson.json: original English, a faithful and natural Vietnamese translation with full diacritics, and IPA in one accent.
  2. For reader and vocab, it makes simple ink stick-figure sketches, the same character in every picture.
  3. A different model or a person checks the English, Vietnamese, IPA and word classes before the build (review gate).
  4. The voices are generated, one of three bundled lo-fi beds is mixed under them at -14 LUFS, frames are checked and the MP4 is rendered.

A run takes about 15 minutes plus one review. Lesson text is always written for the video; it never copies an app, a book, a song or another creator.

Voices

The default voices are free and offline: English with Kokoro (voice af_heart, slowed slightly for learners) and Vietnamese with VieNeu (voice Hải Đăng). The joke format is voiced in English only. Soniox, ElevenLabs or Gemini voices are used only when you ask.

What you need

  • Claude Code or Codex.
  • Node.js 20 or newer, ffmpeg and ffprobe (full build), ImageMagick 7, Python 3.10 or newer with numpy and scipy, and multix. HyperFrames runs through npx and downloads Chromium once.
  • One-time voice setup: python scripts/setup-vieneu.py (about 1.1 GB). Kokoro installs itself on the first build (a 120 MB model).
  • Reader and vocab sketches are made with scripts/generate-illustration.py, which uses a KEY4U_API_KEY.

Install

Copy the skill folder into your skills folder:

~/.claude/skills/agentvid-motion-english-listen-9x16/

For Codex, use ~/.agents/skills/agentvid-motion-english-listen-9x16/.

Who it is for

Anyone making short English lessons for Vietnamese learners: a story reader, a vocab page, a guess quiz or a one-pun joke board, so the viewer can practise and replay.

Not a fit if you want

  • Mistake-correction videos with a mascot, or long courses.
  • Topics other than learning English.

Frequently asked questions

Can I use the videos commercially?

Videos you render are yours for any commercial use. You may not resell, redistribute or share the skill files.

More

Related looks