I tried the most realistic AI -stormmate ever created – if chatgpt or gemini ever get this good is reality in trouble

I’ve spent a lot of time talking to AI. I have tested every voice assistant, every chatbot and any “next generation” conversation ai that tech companies love to hype up. But I have never encountered anything quite like sesame. This AI joint case is not just good, it is creepy exactly to emulate how people speak because of the very imperfections it mimics.

Let’s start with what sesame actually is. Unlike the AI ​​voices we’ve got to know from chatgpt, gemini or go back to the early days of Siri and Alexa, Sesame is designed to act as a human in his failures, not as a perfect customer service agent. AI’s speech is fluent, expressive and unpredictably human. It mirrors briefly when it says something mildly entertaining, hesitates before answering a question and even seems to change his ‘mind’ mid-phrase, break and start a new phrase. Not only does it let me interrupt it, it can also interrupt me and will even apologize for having done it.

(Image Credit: Sesame)

The secret sauce is Sesame’s Conversational Speech Model (CSM) that mixes text and sound for a single process, which means it doesn’t just generate a phrase and then “Read it out.” Instead, it creates speech in a way that reflects how people actually speak, with breaks, UM’s, tone changes and all. Chatgpt and Gemini’s voting options, although impressive, still work in a structured way, generate text and then convert it to speech. Sesame, on the other hand, speaks as if it thinks, which makes its answers feel incredibly natural.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top