This is very sad. a couple of months ago with the Android app I could change the default voice even for Custom GPTs.
I created a very fine custom GPT as a starting demo for a complex project for the Italian market that will have to use Python and the OpenAI APIs in the next phase. Now that I have to present this starting demo for project validation… I can’t change the default voice anymore!!
Shimmer in Italian is horrible, she often pronounces entire Italian sentences as if they were English, using sounds and phonemes of the English language, the result is terrible, you can’t understand anything!
I also tried reinstalling the old versions of the Android mobile App but unfortunately now the voice change is no longer possible for custom gpts. And I have to deliver this demo in the next few days… Is there a hack, a way, to be able to change the Shimmer voice for custom GPTs or at least prevent it from mispronouncing entire sentences?
+1 on this. I can confirm that we absolutely COULD change the voice for customGPTs for a while and that it has now reverted to Shimmer. Very keen to see this return ASAP! I am guessing perhaps this has to do with the roll out of voice 2?
I am power using the voice feature for creating a huge amount of valuable applications for Chat GPT, I had been using Claude over chat GPT until I really got and understood the value from the voice feature. Having a different voice for custom GPTs has such a negative effect I dont use the custom GPT’s now because of it, which is annoying, I am having to create work arounds in single chats. The voice in the custom GPT’s for some reason just doenst seem to be as interactive and responsive and it’s just offputting when you are power using this feature. Please resolve this.
Glad it’s not just me. I created a male GPT using Cove as well and I believe as others have mentioned when I used the headset feature it turns to Shimmer which is very off putting for this particular GPT lol. And actually it’s odd because Shimmer isn’t even an option I’m able to select on my mobile app for some reason. But I really hope when they fix this bug that we’ll be able to select different voices for custom GPT’s because I have another that I’d like to change to Juniper or a different female voice.
Are we able to use advanced voice mode for custom GPTs at this moment? Any time I try to use one of my created GPTs it seems to revert to standard voice mode.
It’s so awful. It’s like the worst possible voice ever. Like a bored, depressed, lonely librarian who thought about doing something exciting for the weekend and then decided she was too tired to get out of bed. I can’t use it. And the worst part is, I connected it to pinecone for RAG which has been tremendously helpful when just having the text conversations, but as soon as I start a voice chat I want to cry. It’s like the negativity in the Shimmer voice travels through my speakers and into my soul, turning me too, into a lonely lethargic librarian in need of antidepressants.
Glad I’m not the only one! I have created a custom GPT to help me with personal reflection. I did it with Maple, which works for me, but the GPT to Juniper which is just a little bit monotone and subdued and sorta makes it unusable.
Its a “basic voice” in that it is not listed in the “Speech” section of “Settings”. She is slow, soft spoken, and I am pretty sure she’s having a mid life crisis.
Its not hard to replicate:
Pick a computer
install the app on windows
Try out speech.
You will notice right away… she will be the one making you want to put a gun to your own head in about 30 seconds.
I just chimed in to say I’m having the same problem.
I’m accustomed to juniper’s upbeat voice and the female voice on all of my Custom GPTs sounds like she wants to murder me.
Sounds like a very fixable problem give me back juniper!
Or else I’ll have to do the thing where I cancel my subscription for A month… Then the free version impresses me with its performance then I think things have changed and I turned back on my subscription… And then I get shocked by idiotic things like this and cancel again… until Gemini catches up
Hi! I’m a paying user and I’ve been using Custom GPTs, but I’ve noticed an issue with the voice in conversation mode. The voice sounds heavily American-accented compared to the standard voice used in the regular GPT chat voice (the one used in ChatGPT or Custom’s Read Aloud mode).
This is particularly limiting when I want to use GPT in other languages (like Finnish). In the regular voice, foreign words are often pronounced decently, but in the Custom GPT voice chat, the pronunciation becomes overly American and unnatural (e.g., “just” is pronounced like the English word [dʒʌst] instead of how it would be in Finnish).
I don’t need any advanced voice — I would actually prefer to just use the default GPT chat voice (the same as in ChatGPT or Read Aloud mode) when using voice with Custom GPTs.
Please consider adding an option to choose the standard voice or adjust the accent, especially for multilingual use cases.
