audio-producer-agent

これはレビュー時に保存された抜粋です。完全で最新の内容は外部ソースを確認してください。

---
name: audio-producer-agent
description: >
  Use this skill to create single-voice audio content like audiobooks, voiceovers, narrations, jingles, and audio ads.
  Triggers: "create audiobook", "generate voiceover", "narration", "audio ad", "radio ad",
  "jingle", "brand audio", "sonic logo", "text to audio", "read this aloud", "audio guide",
  "meditation audio", "soundscape"
  Orchestrates: narration/TTS, background music, and audio assembly.
  NOTE: For conversations/dialogues, use podcast-producer instead.
---

# Audio Producer

Create single-speaker audio content: audiobooks, voiceovers, narrations, jingles, and more.

**This is an orchestrator skill** that combines:
- Text-to-speech / narration (Gemini TTS, ElevenLabs, or OpenAI TTS)
- Background music / ambient audio (Lyria)
- Audio assembly (FFmpeg via media-utils)

**For dialogues and conversations**, use `podcast-producer` instead.

## What You Can Create

| Type | Example |
|------|---------|
| Audiobook | Long-form narration of text/chapters |
| Voiceover | Narration for video, presentation, or slideshow |
| Audio ad | Radio or podcast advertisement |
| Jingle | Short brand music with optional tagline |
| Sonic logo | Audio brand identifier (few seconds) |
| Audio guide | Museum/tour style narration |
| Meditation | Guided relaxation with ambient audio |
| Soundscape | Ambient audio environment |

## Prerequisites

- `GOOGLE_API_KEY` - For Gemini TTS (voice) and Lyria (music)
- FFmpeg installed: `brew install ffmpeg`

## Workflow

### Step 1: Gather Requirements (REQUIRED)

⚠️ **DO NOT skip this step. Use interactive questioning — ask ONE question at a time.**

#### Question Flow

⚠️ **Use the `AskUserQuestion` tool for each question below.** Do not just print questions in your response — use the tool to create interactive prompts with the options shown.

**Q1: Type**
> "I'll create that audio for you! First — **what type of audio?**
> 
> - Audiobook / narration
> - Voiceover (for video/presentation)
> - Audio ad / radio ad
> - Jingle / sonic logo
> - Meditation / guided audio
> - Or describe your own"

*Wait for response.*

**Q2: Content**
> "What's the **text/content** to speak?
> 
> - Paste the text here
> - Or describe what you need and I'll write it"

*Wait for response.*

**Q3: Voice**
> "What **voice style**?
> 
> - Professional
> - Warm/friendly
> - Energetic
> - Calm/soothing
> - Dramatic
> - Or describe your own"

*Wait for response.*

**Q4: Music**
> "Do you want **background music**?
> 
> - Yes — describe the style (ambient, upbeat, cinematic, etc.)
> - No — voice only"

*Wait for response.*

**Q5: Duration**
> "What's the **target duration**?
> 
> - Let it be natural length
> - Or specify (e.g., 30 seconds, 2 minutes)"

*Wait for response.*

#### Quick Reference

| Question | Determines |
|----------|------------|
| Type | Processing approach and output format |
| Content | TTS input text |
| Voice | Voice selection and style parameters |
| Music | Whether to generate and mix music |
| Duration | Pacing and content length |

---

### Step 2: Prepare the Content

**For narration/voiceover:**
- Optimize text for speech (spell out numbers if needed)
- Add natural pause points (commas, periods)
- Break long content into chunks if > 32k tokens

**For jingles/audio ads:**
- Write the tagline/copy
- Determine music style
- Plan structure: music intro → voice → music outro

**For audiobooks:**
- Split into chapters
- Consider different voice styles for different sections
- Plan ambient music (subtle, low volume)

---

### Step 3: Generate Assets

#### Type: Voiceover / Narration

**Generate narration (Gemini TTS):**
```bash
python3 ${CLAUDE_PLUGIN_ROOT}/skills/voice-generation/scripts/gemini_tts.py \
  --text "Your narration text here..." \
  --voice Charon \
  --style "Professional, measured pace, warm and authoritative"
```

**Generate background music if needed (Lyria):**
```bash
python3 ${CLAUDE_PLUGIN_ROOT}/skills/music-generation/scripts/lyria.py \