factual

Voice-to-voice models hallucinate about eight times more

Voice-to-voice models currently have a hallucination rate roughly 8x higher than text because voice generates far more tokens per sentence, and more tokens create more opportunities for errors, which blocks direct enterprise use.

factualpending

Speaker

Jesse

Evidence Quote

“The more tokens you have, the easier it is for something to go wrong. And so the hallucination rate has so far just been a lot higher.”

Source

Jesse Zhang - Building Decagon - [Invest Like the Best, EP.443]— Invest Like the Best
Created: 11/17/2026, 9:59:28 AM

Your notes

Loading notes...