tts-mcp-server-skill_skill

Overview

This skill provides practical guidance for using TTS-MCP-SERVER with ElevenLabs eleven_v3 to produce sultry, seductive, ominous, and emotionally expressive voice output. It focuses on audio tag usage, voice selection, tag-stacking techniques, and script patterns for persuasive or accountability-style prompts. Use it to generate voice announcements, emotional cues, and troubleshoot TTS integration issues.

How this skill works

The skill inspects speak_text calls and voice output requests routed to the ElevenLabs eleven_v3 model and recommends tag stacks, stability settings, and voice IDs for the desired mood. It supplies ready-to-use script patterns and non-verbal cues, plus configuration hints for running the TTS MCP server and managing voice rotation to avoid habituation. It also provides troubleshooting steps when tags are ignored or output sounds robotic.

When to use it

• Generating voice announcements with strong emotional character (sultry, ominous, mischievous).
• Applying audio tags for expressive performance or timed accountability prompts.
• Crafting ‘naughty professional’ scripts for focus, denial, or reward scenarios.
• Integrating ElevenLabs eleven_v3 into automated workflows or voice-mode agents.
• Troubleshooting tag responsiveness, pitch/stability, or voice selection issues.

Best practices

• Start scripts with an ellipsis (…) to force an inhale and prevent first-syllable cutoff.
• Stack 1–3 complementary tags (texture + emotion) for complex directions, e.g. [Close to mic][Sultry][Commanding].
• Use stability settings around 0.35–0.40 for balanced expressiveness without instability.
• Rotate voices (use each voice ~3 times) to prevent listener habituation.
• Match tags to a voice’s natural character and reduce stability if tags appear ignored.

Example use cases

• Focus timer announcements that alternate seductive control and hard stop prompts.
• Automated accountability partner scripts that deny and reward to enforce breaks.
• Ominous mid-work check-ins with whisper + exhale for dramatic effect.
• Emergency interrupts using aggressive tags and higher energy settings.
• Voice-mode agents that must switch to text fallback when voice is disabled.