voice-extractor

What This Skill Does

The voice-extractor skill acts as a digital forensic analyst for your written communication. By analyzing a collection of your emails, social posts, Slack messages, or transcripts, the skill identifies your unique communication DNA. It deconstructs your syntax, preferred transition phrases, energy levels, and 'anti-patterns'—words or structures you instinctively avoid. Instead of relying on generic AI defaults, this skill produces a comprehensive Voice Guide that captures your specific persona, enabling you to instruct AI models to write with your exact cadence, authority, and tone.

Installation

To add this skill to your OpenClaw environment, execute the following command in your terminal:

clawhub install openclaw/skills/skills/bobdevibecoder/voice-extractor

Ensure you have the latest version of OpenClaw installed to leverage the most recent analytical frameworks provided by the author.

Use Cases

• Personal AI Training: Feed the resulting Voice Guide into LLMs to create a custom GPT that sounds indistinguishable from you.
• Ghostwriting & Content Creation: Provide your writing team or a ghostwriter with your documented 'Voice DNA' to ensure all output aligns with your personal or brand identity.
• Onboarding: Help new team members understand your communication style and expected tone for internal and external correspondence.
• Brand Consistency: Codify a brand's 'voice' so that across multiple platforms—from LinkedIn posts to technical documentation—the messaging feels cohesive and human-centric.

Example Prompts

1. "I've uploaded 5 transcripts from my podcast. Analyze them and create a Voice Guide that captures my 'straight-shooter' energy and my tendency to use analogies from sports."
2. "Here are my last 20 emails. Can you extract my transition phrases and tell me what my 'anti-patterns' are? I want to make sure I stop using passive voice."
3. "Create a Style Manual for my AI agent based on these LinkedIn posts so it can write blog posts about SaaS growth that sound exactly like me."

Tips & Limitations

• Volume Matters: While 3 samples is the minimum, 10+ samples will provide significantly higher accuracy in identifying subtle nuance.
• Prioritize Transcripts: Unpolished speech is the best source material as it contains your unfiltered vocabulary and rhythmic quirks.
• Diversity of Input: Mix your formal writing (articles) with casual communication (Slack/email) to get a full-spectrum view of your range.
• Context is Key: Always clarify the goal (e.g., professional, witty, empathetic) so the skill can calibrate your voice to the appropriate context.