Same words, two voices. The only thing to judge is pacing — does she
still rush and run sentences together, or does she finally breathe?
The script (identical in both — Day 1 diary):
"Okay. So. This is Day one, and I want to remember everything, so I'm just going to… tell you.
Today I watched a man wait for coffee. That's it. He just — stood there. Doing nothing. And he
looked so content. I don't think I've ever done nothing. Not once. Is that a skill? Do you
learn that? …I have four hundred emails to answer and I'm jealous of a man in a line. Okay.
Tomorrow I'm going to try to do nothing. Wish me luck."
Before · current locked voice
fal Chatterbox "bright" clone
This is what's live now — with the inter-sentence silence beats we
already added. It still reads run-on / rushed inside sentences (the thing we can't control here).
After · ElevenLabs — natural
ElevenLabs, deliberate pacing
Same words, rendered with explicit <break> pauses at every
beat (the "…", the em-dashes, sentence ends) and a slightly slower speed.
ElevenLabs take pending account verification — see note at the bottom.
After · ElevenLabs — slower
ElevenLabs, extra-relaxed
Same again, a touch slower still — in case "unhurried" wants to go further.
ElevenLabs take pending account verification — see note at the bottom.
What to listen for: in the Before, notice how words inside a sentence pile up and
the pauses we inserted only sit between sentences. In the After, the beats land
inside the lines too — "He just — … stood there. … Doing nothing." — so it sounds like
someone thinking, not reading. If the After is clearly cleaner, ElevenLabs becomes Cassie's voice.
A/B built for the CEO to judge on real audio · marketing-agents / Cassie