Home » AI » Forget ChatGPT and Gemini—Sesame AI’s Voice Feels Eerily Human!

Forget ChatGPT and Gemini—Sesame AI’s Voice Feels Eerily Human!

by Ravi Teja KNTS
0 comment

Forget ChatGPT and Gemini. Sesame AI’s realistic voice technology is dividing and amazing the internet, and for good reason. We’ve all seen AI voice assistants from tech giants—Siri, Alexa, ChatGPT’s Advanced Voice Mode, and Google’s Gemini Live. They promised a lot but often delivered robotic, flat responses. If you have seen OpenAI’s Demo and their Voice mode, you will know the difference I am talking about. But then came Sesame, using it felt like talking to a real person. Here’s everything to know about and how you can try it out. 

What is Sesame AI?

Sesame is just another AI voice assistant similar to Siri, ChatGPT, and Gemini Live. Founded by Oculus co-founders Brendan Iribe and Ankit Kumar, Sesame AI offers something refreshing and scary good.

These are not your typical chatbots or voice assistants. Sesame doesn’t even call them assistants. Instead, they are “conversationalists” and “voice companions.” The idea is not just to respond but to engage you with a voice that sounds almost human. And not just in how it talks but in how it listens, pauses, and reacts.

There are two companions currently – Maya (female) and Miles (male).

Our Experience Using It

When I first tried Sesame’s Maya voice companion, I was genuinely surprised. It’s fast to respond and most importantly sometimes it doesn’t just wait for its turn to talk—it can interrupt you, apologize if needed, and handle messy, back-and-forth dialogue smoothly. What? It was more than just impressive. Unlike ChatGPT’s Advanced Voice Mode, which often sounds too polished, or Gemini Live’s robotic TTS engine, Sesame’s companions had imperfections that made it feel humane.

To test, I asked Sesame to guess my favorite movie genre and also kind of mentioned that I like horror directly in the question itself. The AI playfully responded, ‘Well, you’ve already let the cat out of the bag—looks like you’re a horror fan!’ It wasn’t just the words but the tone and timing that made it feel like a genuine, human reaction. It also remembered the past conversation and understood all the callbacks. 

Also, unlike regular voice modes from other AIs, these “conversationalists” and “voice companions” AIs bring up new questions and ask you in return which is quite weird explaining things to AI. Even when I asked for birthday party ideas, it didn’t just throw random suggestions. It asked what my friend liked, adapted to my budget, and offered creative ideas that felt tailored to my situation. It was so good and natural that it was a little unsettling. Some users online even admitted they felt emotionally attached to it, and one person’s 4-year-old cried when they couldn’t talk to Maya again!

Also Read:

However, there are some limitations too. For example, sometimes it messes up its reasoning and cannot figure out things that a normal human can easily point out. Also, it was not able to identify my emotion through tone, nor my gender or ethnicity.  Though it has a few limitations, it’s much better to acknowledge that and make it feel natural.

Sesame’s technology is backed by a Transformer-based Conversational Speech Model (CSM), trained on nearly 1 million hours of audio. It doesn’t follow the traditional two-step text-to-speech approach. Instead, it merges text and audio processing into a single, fluid experience, allowing it to sound like it’s thinking while speaking. It’s not full-duplex—it doesn’t process your speech until you finish—but it feels like it is. And that’s the magic. However, the company is planning to go full duplex in the future. 

How You Can Try It Right Now

If you’re curious, you can try Sesame’s Maya and Miles for free. It’s recommended to use Google Chrome for the best experience.

Just head over to Sesame’s website and start talking to them in the Demo section. Whether you find it amazing or a bit too realistic, one thing is for sure—Sesame AI has set a new benchmark for what a voice companion can be.

The future of AI voices is here, and it might just make you question who—or what—you’re talking to.

You may also like