i saw something about sky but going back to this issue i tried switching voices and still get the same thingy where the sound icon is not reacting when i click. did u try the cors workarounds on the firefox browser
I’ve tried changing voices as well, it doesn’t matter. I don’t understand what you mean by the “the cors workaround”. Could you please elaborate?
when u open chatgpt in a windows sandbox did u try opening the console in edge? open it, press f12 and click on open dev tools. then u can see errors in the console. try generating some text and clicking on the sound to see if there are errors. also try doing it twice by pressing ctrl+shift+r while console is open to see fresh entries. i was playing with it in firefox, trying to disable extensions or add new ones like cors everywhere and laboratory to allow permissive content security policy (csp)
Same issue here, It’s really frustrating for those who are using this feature in their workflow.
You mean this?
I understand nothing from this though. I see the CORS error I think you’re referring to.
I would have no clue what to do about this. I’m no web developer expert and don’t understand any of this.
The only thing I do understand is that I get the same issue, read aloud not working, regardless if I run in Windows Sandbox, in either Chrome, Edge or Firefox, or my regular browser on my work computer, or any browser on a mobile device or even on my home computer.
- 4 different physical devices (1 work computer, 1 home computer, 1 private phone, 1 work phone) using Edge, Firefox and Chrome, except Safari on the phones.
- Windows Sandbox, with completely clean installations of Edge, Firefox and Chrome
- Safari on iPhone
All result in the same: “read aloud” doesn’t work, except for the iPhone app that I’ve tested.
I asked ChatGPT where I had the following conversation:
So it sounds like it doesn’t matter what happens on the client side. OpenAI has some fixing to do on their end. Whether this is the cause for “read aloud” not working though, I of course have no idea.
Btw @ISnaked
The TTS Api is actually working – tried node and curl and it probably would take them only 3min to fix the frontend.
The quality of the voices available through the API aren’t as good as the “read aloud” ones. They seem to work very different. I use Juniper in the “read aloud” feature a lot, as I like the tonality and emphasis I’m able to get with that voice. Or have you been able to use the “read aloud” voices using an API? I’ve looked for this but haven’t found anything on it.
I am experiencing an issue with the “Read Aloud” feature, which has stopped working on both my phone and laptop. This feature was working fine on both devices about 4-5 days ago, and I have not changed any configurations since then. I am currently on a paid subscription.
You’re saying that “read aloud” isn’t working in your mobile app?
Much to my frustration, I’ve also had this. Really hope they get on the problem soon as its a heavy use feature of mine now as I tend to work while I listen to the responses. I’m so much slower at the moment.
Yes it has stopped working on both my phone and laptop
This is a first. The reports so far has been about “read aloud” stopping to work in web browsers, but not in any of the mobile apps. I just tested, and “read aloud” still works in my iphone app.
I can confirm this. The Read Aloud feature hasn’t been working properly since the GPT-4o update.
I hope it gets fixed soon🙂
It’s shocked such easily fixed stuff being postponed for so long
I suspect it has more to do with the shift to redirecting https://chat.openai.com to https://chatgpt.com while the page itself is still trying to load assets from the old domain, possibly causing at least some of the CORS issues.
Because it very much appears to me to be as simple as the JavaScript file for that feature not getting loaded.
If I wasn’t a few hundred miles from home traveling solo with a two-year-old right I’d try to dig into it further, maybe tonight when we’re home I’ll look into it further if it hasn’t been resolved by then.
@anon22939549
That seems very likely.
Please take care and have a safe trip back!![]()
For a general explanation about CORS, here is a helpful site: Cross-Origin Resource Sharing (CORS) - HTTP | MDN
The issue might be caused by the recent change from the “chat.openai.com” domain to the “chatgpt.com” domain, which prevents JavaScript from the original domain from being referenced. Either way, OpenAI needs to fix this.
Yes, this sounds like a very plausible explanation. This should generate a lot of errors in a log system somewhere. (Should be millions of rows just related to this. At least, one would hope that is the case)
Same problem. Chrome browser. Windows 10.
For me, not only is the “Read Aloud” function not working correctly, but the interface has also not been updated. Even with access to GPT-4o, the new interface has not been updated for me yet. This is probably related to the domain change.
Is this related to “OpenAI halts use of voice over Scarlett Johansson”?
Bring back read aloud even if it doesn’t come with Scarlett’s voice, no one cares about it and we just want the function back.
good to know it’s not just me. question is, what can I do about this, and has the OpenAI team picked up on this?




