podcast-generator

name: podcast-generator description: Converts blog posts into podcast audio using Kokoro TTS engine with multi-language support license: MIT compatibility: opencode metadata: audience: blog-contributors workflow: content-distribution

Podcast Generator Skill

Use this skill when converting blog posts to podcast audio or generating TTS-optimized transcripts for enhanced audio quality.

What I Do

• Creates TTS-optimized transcripts for better audio quality
• Generates podcast audio files using Kokoro TTS engine
• Supports both English and Traditional Chinese posts
• Automatically updates RSS feeds for podcast distribution
• Manages multi-language podcast feeds

When to Use Me

Use this skill when you need to:

• Convert blog posts to podcast audio
• Generate TTS-optimized transcripts
• Create multilingual podcast content
• Update podcast feeds
• Improve audio quality for specific posts
• Troubleshoot podcast generation issues

Key Features

• Multi-language support: Automatic English and Chinese processing
• TTS optimization: Create transcripts optimized for natural speech
• Transcript management: Support for language-specific and generic transcripts
• RSS feed generation: Automatic feed updates for podcast distribution
• Auto chunking: Handles long posts with automatic audio chunking
• Voice configuration: Customizable voices for different languages

Quick Start

Basic Generation

uv run podcast-generate --posts "my-post-slug"

Force Regeneration

uv run podcast-generate --posts "my-post-slug" --force

All Posts

uv run podcast-generate --all

Transcript Management

File Organization

Place transcripts in: src/content/blog/transcripts/

Naming conventions:

• slug.en.txt - English transcript (preferred)
• slug.zh-hant.txt - Chinese transcript (preferred)
• slug.txt - Generic transcript (fallback)

Transcript Priority

1. Language-specific (if exists)
2. Generic (if exists)
3. Auto-derived from post markdown

Creating Transcripts

Convert blog post markdown to spoken text:

• Remove markdown syntax (headers, lists, code fences)
• Describe code blocks in plain language
• Expand abbreviations ("e.g." → "for example")
• Add natural punctuation for pauses
• Keep sentences short for better pacing

Output Structure

public/podcasts/
├── feed.xml                    # Main feed (all languages)
├── post-slug.mp3              # English audio
├── post-slug.zh-hant.mp3      # Chinese audio
├── en/
│   └── feed.xml               # English-specific feed
└── zh-hant/
    └── feed.xml               # Chinese-specific feed

Language Support

English:

• Voice: af_sarah
• Model: Kokoro v1.0
• Auto-detected from lang: "en"

Traditional Chinese:

• Default voice: zf_001 (female)
• Alternative: zm_009 (male)
• Model: Kokoro v1.1-zh
• Auto-detected from lang: "zh-hant"
• 90+ voice options available

Verification Commands

# Check audio file
ls -lh public/podcasts/post-slug*.mp3

# Verify duration
ffprobe public/podcasts/post-slug.mp3 2>&1 | grep Duration

# Check RSS feeds
grep "post-slug" public/podcasts/feed.xml
grep "post-slug" public/podcasts/en/feed.xml
grep "post-slug" public/podcasts/zh-hant/feed.xml

Troubleshooting

"Found 0 posts": Verify frontmatter YAML is valid and pubDatetime is in the past

Audio not updating: Delete existing file and use --force flag

Model download fails: Check disk space and internet connection, retry download

FFmpeg errors: Verify FFmpeg is installed and in PATH

Prerequisites

• Python environment: uv sync
• FFmpeg installed
• Valid post frontmatter with proper pubDatetime
• Write permissions to public/podcasts/

Quality Tips

• Create transcripts for complex/technical posts
• Use language-specific transcripts for better pronunciation
• Listen to generated audio for quality assessment
• Test that RSS feeds validate correctly
• Monitor podcast subscriber feedback

Integration

Works seamlessly with:

• Blog Writer (content creation)
• Translator (multilingual content)
• Fact Checker (content verification)
• GitHub Publisher (deployment)