Same problem as all of you. Have stopped using it due to the interruptions. Bring back the manual hold button, it will not take anything away from your slick design.
1 Like
Thatās really frustrating, especially for such a great feature. I hope they take these comments into account and fix it soon 
Same here. Uffff. So frustrating. I arrived here searching on Google āchat gpt advanced interrupt conversationā. Yesterday I payed 22ā¬ to fully enjoy advanced voice mode, yesterday I was telling the World (family and friends, in social media too) how Cool was Advanced voice mode after I tried it with the free account. Then now Iām so frustrated realising it doesnāt work 
.
Yeah, adding another voice to the pile here. This is a massive issue. I bought Plus just to use advanced voice mode and I got so frustrated within 3 minutes that I havenāt tried it again.
A real conversation partner knows when you are pausing mid-sentence to think and when you finish your thought. Iām not exactly sure how advanced voice mode works, but if itās just transcribing your words into text and feeding that into an LLM, thereās no way it can simulate social awareness. Humans rely on tone of voice, pacing, and body language to infer the right time to speak.
If itās not possible for ChatGPT to accurately guess the appropriate time to speak, there needs to be a temporary solution for this, like touching and holding the screen until you want ChatGPT to respond. Iām dumfounded they got rid of that feature. BRING IT BACK IMMEDIATELY!
On a psychological level, when ChatGPT interrupts me, my social brain goes āwow, this person has no social awareness. They are rude and annoying. Are they even listening? What is wrong with them!ā Especially since the chatbot is simulating human voice and expression, my brain wants to judge them on human terms. So the chatbot comes across as an extremely poor listener.
Iām human, and I donāt like talking to people with no social awareness and poor listening skills. That sums my user āexperienceā with ChatGPT advanced voice.
Iād bet money several product designers at OpenAI recognized this problem and proposed keeping the āpress and holdā feature in the new voice mode. Some higher-ups probably instructed them to remove it because the button implies that the new voice mode isnāt as responsive or natural as they marketed it to be.
My advice to OpenAI - donāt be like Apple. Donāt remove useful features and pretend like weāre āin the future.ā
PLEASE bring back the āpress and holdā feature ASAP or Iām going back to Claude.
1 Like
Does OAI even read these messages? I canāt believe this issue has been outstanding for months with no resolution. Itās completely useless right now. No amount of instruction will get it to stop interrupting. How is this not a major priority for this feature right now?
Not only people with ADHD but with depression as wellā¦ the older voice model has been an unexpected source of support for me, especially in dealing with my depression. Its calm and steady nature has a soothing effect on my thoughts, helping me feel more grounded during our interactions.
The modelās ability to ask valuable and reflective questions creates a space for deep thinking, which is incredibly important for processing my emotions. I also appreciate the way it never rushes through a topic, allowing the conversation to flow naturally. The long pauses it includes give me time to focus on my thoughts and work through them at my own pace.
Additionally, the consistent follow-up questions help me explore ideas and emotions further, making me feel genuinely heard and supported. These qualities make the older voice model feel less like a machine and more like a calm, patient presence that I can rely on.
I truly hope these aspects can be preserved or reintroduced in future updates. They make a meaningful difference for users like me who find comfort and emotional clarity in these interactions.
I am taking a photo and sharing with chat , that seems easier, thank you!
I donāt understand why the feauture from regular voice mode, where you hold the button in the middle while youāre still speaking, doesnāt exist in Advanced Voice Mode. This is the one thing that really keeps me from using AVM; the constant interruptions even if Iām silent for just two seconds.
1 Like
+1 to whatās been said here. The ideal solution would be a user-set parameter that determined how long of a pause to allow before AVM responds. In some cases I am fine with the pacing, but when Iām thinking through something tricky I would like to set response speed to ārelaxedā or āslowā rather than ābriskā which is currently the default. Regular voice mode still allows this but removes the hands off benefits of OVM.
I just wanted to add to this thread. I would love for more firm control of ChatGPTās behaviour in advanced voice mode. Every interruption causes me to either cease my train of thought or to talk over the GPT response, which feels unpleasant. Please fix this!
Iāve been having a similar but slightly different problem with advanced voice mode. This is when using the iOS ChatGPT app on my iPhone 16.
When using advanced voice mode, I am able to complete speaking my prompt without interruption but when the app starts its response, it interrupts itself as if Iāve said something when Iāve said nothing at all. I have experienced this while sitting in a quiet room with no ambient noise at all using the iPhoneās primary mic, and Iāve even tried while using wired headphones to rule out the iPhoneās mic being too sensitive.
Sometimes the text transcription logs these interruptions with what seems like random text, but theyāre almost always words I havenāt actually spoken. And other times it just registers as 1 second of no text - when that happens, I usually see more than one entry like that.
It is definitely frustrating and wasnāt happening when I first got access to advanced voice mode. I have made sure the ChatGPT is up to date, Iāve reinstalled it to make sure of that. And Iāve confirmed my mic settings are correct for the app. I have no issues with voice commands using Siri or any other apps that use my phoneās microphone to convert speech into actions.
Hoping OpenAI actually pays attention to these threads, or at least has an AI model culling the comments to find opportunities to improve the app.
1 Like
I just wanted to post this verification that on my system if you start your chat off in text mode and then switch over to voice mode you will get the non-advanced mode where you can hold the center of the screen To keep from being interrupted.
1 Like
Same here. Not an ADHD but I drive a lot and talk to chatGPT.
It is soooooo annoying! Even if I pause for one second it starts talking.
I tried to tell it than only when I explicitly use some magic work, such as āoverā can it talk, it agrees but never keep the promise.
I will unsubscribe, it is aweful
This looks promising. It looks exactly what us slow talkers need to have as an option.
Let me know if you need a beta tester.
Its always something. Why is this not fixed yet? Its so aggravating. It should be programmed to understand that humans need time to think. With all the other issues as well, this one is almost top of the pile. Just cancelled my subscription. off to Deepseek we go.
Simply amazing, that still, A YEAR LATER, OpenAI have not either seen this, or are too incompetent to solve it. I doubt the latter is the case, and if it is, it would be one of the biggest oxymorons in history. It should take a junior dev about an hour, at most, to solve.
I think there is a way to edit this in the settings
Iām like a year late logging back on to these forums. Itās good to know Iām not the only one with the struggles lol. Thatās hilarious about the violations, cause Iāve got them too. I think by now theyāve managed to bring it back somewhat. Cause it looks like that same blue/white wave animated circle (opposed to the old plain white one) but you can still hold it. This past months they changed it to a button in the chat itsel but you can change it back in the settings.
Anyway, hope all is going well with you. This has been a game changer for my adhd to work through so many thoughts. I wish I had it years ago. But glad I have it now.
Yeah dude, Iāve definitely experienced that. I think hitting by mute may help? (Iām not 100% sure).
Actually an update for this whole thread, I just realized I donāt even use the white circle feature anymore. I donāt even use a voice note at all. Because it does things like that and then it also like completely cancels, my voice message midway through talking. I donāt know why it does this all the time. So Iāve just reverted to voice to text and I didnāt think I would get used to that ever but itās been such a lifesaver. Plus, I can control the responses whenever ChatGPT replies because I can just be able to rewind at my own leisure and what not. so itās actually better but itās just not as quick responding, and it doesnāt feel as natural (at first) but once you get used to it, it feels just the same in my opinion and it does feel natural after a while.
For me itās like a walkie talkie lol. You just have to click a button and talk (voice to text) and then let it answer, and press the voice button so it speaks out loud. It just takes habit and repetition and enjoy being able to pause and rewind if needed.
1 Like
Great suggestion, I had not considered trying voice to text so itās not as much an open conversation. Not as convenient but far less frustrating if it eliminates the interrupted responses. 
Itās still disappointing that I cannot have a conversation with the model as intended. I am typically using the advanced voice feature at home with no other ambient noise present, so it feels like I am using the feature under optimal conditions yet it still somehow stops and starts. Occasionally when I check the transcription, it will capture me saying words completely unrelated to the question I asked and which I 100% did not say. It feels like the model is interpreting even the tiniest ambient noises as communication, so it feels like OpenAI needs to dial that back a bit.