---

name: SSML Script Generation tier: 3 load_policy: task-specific description: Generate production-ready SSML scripts for Google Cloud TTS version: 1.0.0 parent_skill: production-operations

SSML Generation Skill

The Script Is the Blueprint

This skill handles the technical generation of SSML scripts that will be converted to voice audio.

---

Purpose

Transform journey concepts into properly-formatted SSML that produces natural, hypnotic voice output.

---

Critical Rules

Voice Rate: ALWAYS 1.0

Copy<!-- CORRECT: Always use rate="1.0" -->
<prosody rate="1.0" pitch="-2st">
  Text content here... <break time="2s"/>
</prosody>

<!-- WRONG: Never slow the rate -->
<prosody rate="0.85" pitch="-2st">
  This sounds robotic with Neural2 voices
</prosody>

Why: Google TTS Neural2 voices sound unnatural at reduced rates. Achieve pacing through <break> tags instead.

---

Speech Profiles by Section

---

Break Duration Guidelines

---

Section Header Template

Copy<!-- ═══════════════════════════════════════════════════════════════ -->
<!-- SECTION: [SECTION_NAME] -->
<!-- Duration: [X] minutes | Pitch: [X]st | Purpose: [PURPOSE] -->
<!-- ═══════════════════════════════════════════════════════════════ -->

<prosody rate="1.0" pitch="[X]st">
    [Content...]
</prosody>
<break time="[X]s"/>

---

Sound Effect Markers

Place SFX cues on their own lines in script_production.ssml:

Copy[SFX: Deep ceremonial bell tone, resonant, 4 seconds with natural decay]

<prosody rate="1.0" pitch="-2st">
  The sound fades into silence...
</prosody>

Important: SFX markers are stripped out for script_voice_clean.ssml before TTS generation.

---

Two-File System

Workflow:

1. Write script_production.ssml with all SFX markers
2. Generate script_voice_clean.ssml by stripping [SFX:...] lines
3. Feed script_voice_clean.ssml to TTS

---

Section Duration Targets

---

SSML Validation

Before TTS generation, always validate:

Copypython3 scripts/utilities/validate_ssml.py sessions/{session}/working_files/script.ssml

Common Validation Errors

---

Integration with Tier 1

All SSML content must follow Tier 1 rules:

---

Command Reference

Generate Script

Copy/generate-script {session}

Strip SFX for TTS

Copygrep -v "^\[SFX:" script_production.ssml > script_voice_clean.ssml

Validate

Copypython3 scripts/utilities/validate_ssml.py sessions/{session}/working_files/script.ssml

---

Output Files

---

Related Resources

• Skill: tier1-neural-core/hypnotic-language/ (language patterns)
• Serena Memory: script_production_workflow
• Serena Memory: voice_pacing_guidelines
• Doc: prompts/hypnotic_dreamweaving_instructions.md