VideoAsCode (vac)¶

Transform Marp Presentations and HTML Slide Decks into Videos with AI Voiceovers

vac is a command-line tool that automates the conversion of Marp presentations and native HTML slide decks into professional videos with AI-generated narration.

Features¶

• Parse Marp presentations with voiceover in HTML comments or JSON transcripts
• Record HTML slide decks with narration in data-vo attributes or reveal.js speaker notes — served and recorded in one command, each slide timed to its own narration
• Text-to-speech using ElevenLabs, Deepgram, local F5-TTS, and other providers via OmniVoice
• Browser automation with Rod to display slides and record demos
• Screen recording with synchronized audio using ffmpeg
• Multi-language support with BCP-47 locale codes (en-US, en-GB, fr-CA, etc.)
• Platform-optimized output for YouTube, Udemy, Coursera
• Crossfade transitions between slides
• Subtitle generation from voiceover timing or speech-to-text (Deepgram or local Whisper)
• Course lessons — record a deck as one video per lesson, with a manifest for your course platform, re-recording only the lessons that changed
• Audio caching per voiceover, keyed by text and voice, so only edited narration is sent to TTS again
• Avatar presenter overlay (optional) via OmniAvatar — HeyGen, Tavus, bitHuman
• Local, offline TTS & STT (optional) — F5-TTS + Whisper on Apple Silicon via OmniVoice MLX providers, no API keys or cloud calls

Quick Example¶

# Simple: inline voiceover comments
vac slides video --input slides.md --output video.mp4

# Advanced: multi-language transcript
vac slides video --input slides.md \
           --transcript transcript.json \
           --lang es-ES \
           --output video_spanish.mp4

# Serve and record an HTML deck; narration comes from each slide's data-vo
vac html video --input deck/index.html --output deck.mp4

# Write a transcript (for translations) and a reusable video config
vac html init --input deck/index.html --provider deepgram --voice aura-asteria-en

# One video per lesson (slides with data-lesson start lessons)
vac html video --input deck/index.html --lessons all

# Record browser demo with voiceover
vac browser video --config demo.yaml --output demo.mp4

# Multi-language with audio caching
vac browser video --config demo.yaml --output demo.mp4 \
           --audio-dir ./audio --lang en-US,fr-FR,zh-Hans

# With subtitles burned in
vac browser video --config demo.yaml --output demo.mp4 \
           --subtitles --subtitles-burn

How It Works¶

flowchart LR
    A[Marp MD] --> B[Parse]
    B --> C[TTS Audio]
    C --> D[Record Slides]
    D --> E[Combine]
    E --> F[Video.mp4]

1. Parse - Extract slides and voiceover from Marp markdown
2. Generate Audio - Convert text to speech via ElevenLabs/OmniVoice
3. Render HTML - Use Marp CLI to create HTML presentation
4. Record - Screen capture each slide with audio sync
5. Combine - Concatenate slides with optional transitions

Getting Started¶

Use Cases¶