ChatGPT Voice Update is WILD! artwork

ChatGPT Voice Update is WILD!

AI News Today | Julian Goldie Podcast

July 26, 2026

ChatGPT Voice Update is WILD!
Speakers: Julian Goldie
**Julian Goldie** (0:00)
Open up Google inside my Chrome browser.

**SPEAKER_2** (0:04)
Sure, opening that now.

**Julian Goldie** (0:05)
Can you type hello into Google?

**SPEAKER_2** (0:07)
Typing that in.

**Julian Goldie** (0:08)
Would you be able to increase your speaking speed by like 50%?

**SPEAKER_2** (0:13)
Sure, I'll speed up.

**Julian Goldie** (0:14)
Write a journal inside the Notes app, and just journal about like what you did today, what you learned, et cetera.

**SPEAKER_2** (0:23)
On it, I'll keep it concise. I think a quick daily journal in Notes would be a great way to capture what you did, what you learned, and what's next.

**Julian Goldie** (0:33)
Perfect. Can you see my face right now?

**SPEAKER_2** (0:36)
Let me check.
Not right now. I'm seeing your Notes app, not your camera feed or face.

**Julian Goldie** (0:43)
You can't see me in Descript?

**SPEAKER_2** (0:46)
I'm checking that.
I can see you if Descript is open with its camera preview visible, but right now I'm still seeing Notes. Bring Descript to the front and I'll check.

**Julian Goldie** (0:56)
Can you sing for me?

**SPEAKER_2** (0:59)
La la la, la la la.

**Julian Goldie** (1:04)
Reorganize like my Windows, sir. Everything fits on one screen.

**SPEAKER_2** (1:07)
Got it. I'll arrange the layout.

**Julian Goldie** (1:10)
That will do. You can stop there.

**SPEAKER_2** (1:11)
Okay. Stopping now.

**Julian Goldie** (1:14)
So this is the new ChatGPT voice update in the Desktop app, which means that you can control your computer and run multiple AI agents coding in the background just using your voice. You can also bounce it around like a basketball. Look at this.
Having a great time. And this is the new update. So you can see that it basically, it's kind of sticky as a pet on your desktop, and then you can control it. You can mute it and stop it speaking at any time as well. So you just click these buttons so that it doesn't speak to me whilst I'm explaining this to you. And from here, it's powered by GPT Live. So it can speak, it can listen, it can coordinate work. It's also pretty good because as well, like if you were building your own voice agent, it would actually be pretty expensive to use the real-time API. And then if you run out of credits, you have to go and update and add your billing and everything else. It was super annoying. So how good is it? So far, it's pretty impressive. It's quite fun. It does break at time. So, for example, you probably saw before, it took a screenshot and it's got my camera on the screen right now. So it's easily visible, but it couldn't actually understand that. So there are some things where it's great, and there's some things where it's like not that great. But it can also do pretty cool stuff in the background. And let's test out something else else. So if I log in to a different account, as you can see, we're going to see if it can actually just like do something useful, like posting a tweet for us. So let's try this right now.
Can you post a tweet about the power of ChatGPT Voice for me, please?

**SPEAKER_2** (2:42)
Checking that, I'll draft something tight and show it to you for approval before posting. Something like ChatGPT Voice is like having a collaborator.

**Julian Goldie** (2:53)
Just post it, just go on to X and write it right away.

**SPEAKER_2** (2:57)
I've drafted it for the AI Playbook account. Say publish and I'll post it.

**Julian Goldie** (3:03)
Publish.

**SPEAKER_2** (3:05)
Posting it now.

**Julian Goldie** (3:07)
Done. It's submitted and can you just open it up as well?

**SPEAKER_2** (3:10)
Opening it now. It's open in Chrome.

**Julian Goldie** (3:15)
So you can actually post tweets straight away as you can see here. So it posted this tweet, pulled up, control my browser whilst I was talking to you. The only thing that I don't like about it is it's still a little bit of a delay, but I can understand why, especially if you're using a thinking model. So one thing that actually might be really useful is you just switch the API you're using for this and maybe use something like an instant instead of thinking models. But yeah, it actually works, actually does stuff. It did rearrange my windows before. One thing that was super useful actually was whilst trying to record this video, I needed to record the sound on my desktop, but my setup wasn't correct. So what did it do? It actually configured it properly inside this mixer so that I could record the sound of my computer and also hear the sound of my desktop, both of those things I couldn't do before.

9 more minutes of transcript below

Feed this to your agent

Try it now — copy, paste, done:

curl -H "x-api-key: pt_demo" \
  https://spoken.md/transcripts/1000651996090

Works with Claude, ChatGPT, Cursor, and any agent that makes HTTP calls.

From $0.10 per transcript. No subscription. Credits never expire.

Using your own key:

curl -H "x-api-key: YOUR_KEY" \
  https://spoken.md/transcripts/1000778460046