AI Voice Expression — Paul Dolphin Voice Demonstration
A presentation of voice styles and stability settings — how our AI responds with natural human expression
Each voice below uses the same source narration (Riley Natural via ElevenLabs) but varies two parameters: stability (how consistent vs emotionally varied the delivery is) and style (how expressive the voice acts). Listen to hear the difference and understand how we tune voice for context.
| Tokens exchanged (30 days) | ~3,000,000,000 |
| Bibles worth of language | ~2,873 |
| Per day | 96 Bibles |
| Per hour (24/7) | 4 Bibles |
How Our AI Responds — The Conversational Advantage
Voice Style Demonstrations
Three delivery styles using the same script. Notice how stability and style parameters shape emotional tone.
The following are different takes on a morning introduction — a short daily touch-base message that Riley would send on my behalf to a business partner. Just a way to stay connected. A voice check-in. A natural part of the morning routine.
This is an example of the kind of idea prompt I'd love for my partner to contemplate during their morning routine — something worth thinking about over coffee, on the walk, with the dogs nearby.
The prompt here is this: In thirty days, we exchanged almost three thousand Bibles worth of language with AI. That's three billion tokens of conversation — ideas, code, strategy, training data — moving back and forth between humans and machines.
And that doesn't even include the Claude Code accounts I'm running. That number is JUST from the free, open-source LLMs available to anyone. Just the free tier. If we added the paid tools, the real number is significantly higher. Let that one settle in.