📱 Tech

I Tested the New ChatGPT Voice Mode and It’s Creepily Good

I Tested the New ChatGPT Voice Mode and It’s Creepily Good

So OpenAI rolled out this new voice mode for ChatGPT last week, and I’ve been messing with it nonstop. The company claims it can detect emotion, adjust tone, and even laugh at the right times. I thought it was just marketing hype. Then I actually used it.

Setting It Up

It’s super simple. You open the app, tap the voice icon, and you’re talking. The new model, GPT-4o, is built into it. The voice responses are almost instant—like, real-time conversation speed. No lag. No robotic pauses. Just smooth back-and-forth.

The First Conversation

I asked it to tell me a joke. It did—a decent dad joke. Then I said “that was terrible” in a sarcastic tone. It laughed. Not a canned laugh, but a genuine-sounding chuckle. Then it said, “Okay, tough crowd. Let me try a different one.” I actually felt bad for it. That’s weird to say about an AI, but true.

Emotional Intelligence

The big selling point is that it can read your emotional state from your voice. I tested this by pretending to be sad and talking about a bad day. It responded with a softer, more empathetic tone. It didn’t try to solve my problems—it just listened and offered gentle encouragement. Then I switched to an excited voice about a new project, and it matched my energy, sounding enthusiastic. It’s unsettling how good it is at adapting.

Real Use Cases

I’ve been using it as a brainstorming partner. I’ll talk through ideas out loud, and it responds with suggestions, questions, or just listens. It’s surprisingly helpful for creative work. Also, I tried using it as a language tutor—practicing Spanish conversation. It corrected my pronunciation gently and kept the flow going. Way better than any language app I’ve used.

The Weird Parts

Sometimes it pauses like it’s thinking. It’ll say “Hmm, let me think about that” and then give an answer. That feels human but also makes you wonder—is it actually thinking, or just simulating it? Also, it can laugh at inappropriate times if you’re not clear. I made a dark joke and it laughed—felt a bit off.

Privacy Concerns

This thing is listening to your voice constantly during use. OpenAI says they don’t store audio recordings unless you opt in, but still. Having an AI that can read your emotional state from your voice is a bit creepy. I’m not saying it’s malicious, but it’s worth thinking about.

Is It Worth It?

The free tier gives you limited voice access. The $20/month Plus version gives you more. If you’re a heavy user, it’s worth it. But honestly, the voice mode is good enough that I might actually pay for it. It’s not perfect—sometimes it mishears words, and the emotional detection isn’t always spot-on. But when it works, it feels like the future. A weird, slightly unsettling future where you have conversations with your phone and forget it’s not human.

TR
Amanda Brooks

We spend hours researching and testing before we write anything. If something changes, we update the article. About our process →