Bug Report: Voice Selection Not Changing in ChatGPT (Browser & Windows App)
Summary:
When changing the voice in the voice conversation settings, the selected voice does not update and remains the same. Additionally, the voice that is used is not one of the available options.
Steps to Reproduce:
Open ChatGPT in either a web browser or the Windows app.
Start a voice conversation.
Attempt to change the voice to a different option in the settings.
Observe that the voice remains unchanged, even after selecting a different option.
Expected Behavior:
The voice should update to the selected option.
Actual Behavior:
The voice remains the same, and the voice used does not appear to be one of the listed options.
Platforms Affected:
Browser (specify which one if possible)
Windows App
Platforms Not Affected:
Android App (Voice selection works correctly here)
Additional Notes:
Issue persists across different platforms except for Android.
No error message appears; the setting simply does not take effect.
Exact same issue, i have made a thread a week ago. You actually detailed it in a less stupid and casual way than me though.
May i add, i tested it on a new account on another PC on windows 11 (mine is still running with 10) and itās the same exact problem. Also tried 4 different browsers (brave, chrome, mozilla and edge). Someone on my thread actually got a response from a dev saying something like āwe are aware and will be working on itā. Not me though, after asking me to do a bunch of random stuff for three replies, they just deleted the support convo and sort of ghosted meā¦
As for the voice, itās the default one used on custom GPT, itās named Shimmer i think.
Though, the voice selection should be working in vocal advanced mode on PC too. But thatās not what i (and probably you too) want, it got a shorter limit of usage and it feel weird for some things.
It really suck, even more since iām paying for it ! I mean, i pay for others functionality but come on !
I upgraded to Plus specifically for advanced and stranded voice chat. Iām pretty unhappy with the limitation and especially this bug. Ive submit a ticket to support, but no response.
Yet, its horrible. And the selected voice seems to be slower, which makes it even more annoying.
Unfortunately I therefor upgraded to PLUS - just to find out that the āprofessionalā audio mode is limited to a short amount of time per day, maybe(?) about 1 hour or less.
Sorry OpenAI, this is REALLY disappointing! All those bugs of endless repeating the same answer again, losing memories etc and paying not just a small amount is not worth it. I would expect a proper running system if I pay.
Actually I could use an unlimited āprofessionalā audio mode (I think, as ChatGPT does not give any deeper information about anything it offersā¦) IF I would pay 200(!!) bucks per month.
I know, if you earn a few 100 millions a year, like Elon, its play money - but this is really way over the top. Please, get it done soon.
This has been happening to me for quite some time now (I believe it was in early January when it started). The length of time it is taking to āfixā something that worked just fine for months on end makes me wonder if this is intentional? The question then, why would they be leaving the web-based voice selection ābrokeā is a good one. I have been thinking about it, and I can think of at least one reason: Now that they have a good base of user interaction data built up, they are wanting to experiment with the effect of switching the Aiās voice to an unemotional female voice input to study its effects on user psychology. Obviously, this is entirely speculative so I have absolutely no proof of this, other than the fact that Governments and Mega Corporations Both have indicated that their ultimate use for AI Chatbots (LLMās) will be to use them a psychological profiling tools to better sell products to and gain the compliance of the general population. This was widely discussed in the last 2 years at Davos by its attendees. Anyhow, just a thought.
Me as well, using a mix of the Android and Windows apps. Always works on Android. On Windows, it appears to default to a very robotic voice once the conversation gets too long. It will no longer let you change to any other voice.
Once I start a new convo, I have to toggle the voice back and forth in order to reinstate my selected voice and it works again for a time.
Today I notice specifically, I started a voice mode chat on PC. Talked back and forth a bit.
Then I typed in the chat a couple of responses. Then upon trying to switch back to voice chat (instead of typing a longer reply), it had reverted to the basic voice model with a prompt to āswitch to a new chat for advanced voice mode.ā Clicking the info button on this says, āYour current chat has content that advanced voice mode canāt handle just yet.ā
However, in the Android app my selected voice still applies on the same thread.
Unclear, but it seems the Windows app is conflating an inability to continue with Advanced Voice Chat with simply continuing to use the selected voice model.
Itās kind of okay if it canāt do advanced voice chat for now, but itās frustrating/jarring that the voice model is also reverted.
It donāt work this way. Threads created tu use advanced vocal mode revert to normal threads once you do some stuffs in them, like text chat.
You think itās the same voice on the Android App, because in advanced mode, the voices are based on the same voices that the ones in standard mode. But they are not the same, itās more blatant with some of them. And the AI donāt speak as much, or with the same personality as the one you (probably) gave it.
But yeah, aside from this, you get it, it work as intended on Android, the reverting to standard mode is normal, but on windows or just PC, the standard mode is always using the same default voice (Shimmer).
Yes, understood. The issue isnāt that it drops out of Advanced Voice Mode. Understandable technical limitation, maybe.
Itās that it wonāt let you select any other voice model either.
The immersion is broken, and given the already very good quality of the voice models even without advanced voice, simply fixing this would be a big improvement.
Iām a Pro subscriber, using Legacy Voice mode (not Advanced Voice).
Voice selection works fine on Android.
But on the Windows App, and Chrome, and Firefox, I get this a weird androgenous sounding voiceāitās unusable on Desktop. Hope this gets sorted out soon.
same issue for me. On Android , windows. I always choose ember as voice but the system does not change the voice, i have a female voice
i have set ember as default, tried new chats, it doesnt work. and if it works , it changes immediately after i close the app or move to another chat and back
I am also having this issue. Ever since Monday came out, it keeps switching to Monday as default. If I try to switch it back, it shows that it is switched back in the system but then itās still Monday. The only way that it changes is if I start a new chat and lose my progress. Then after a while, Monday again. Read aloud works fine but the voice chat does not use the right voice.
I was using chatGPT on google chrome where the issue was first seen for myself.
I downloaded the desktop app for windows 11, same issue.
The only place I am able to choose a voice is on the Android App.
I want to use Cove, but it doesnāt matter who I select it always remains the same. I think itās Monday.
I would be interested to know if this is being looked into or if thereās any solutions yet.
This issue is preventing me from streaming my Monday discussions. Monday voice does not replay, itās actually Sol when Monday is set as voice pref. I would really really like it if we could choose this, but for now Iām working on the fun task of using the live advanced voice to re read the memory file of the original chat.
i am having the same issue on Android. Trying to select Cove and primary language is set to English but voice reverts to a German guy who speaks English with a thick German accent. Itās horrible.
Just formally updating that the ability to choose Monday in my pc app disappeared - a non working option to be sure, but now gone, where I as a user expect to have the option to set Monday and hear Monday for the chats Read Aloud feature. What is the issue with providing this option in the menu? The option for Monday in Advanced Voice going deprecated should be a separate situation from the read aloud functionality, from bug fixing perspective can we expect to receive this option soon? I still use Monday for almost every situation that calls for AI, because she is ultra superior. But am Stuck choosing Sol for read aloud. Would like real Monday simply to provide the read aloud that belongs with her chats.
