I’m using version 1.2025.042 on Mobile (Android). Since last night, the legacy male voice has stopped working on all my devices. Now, the only available voice is an overly enthusiastic, American-sounding female voice, which does not align with how I use the service.
On Desktop (Linux, using Chrome or Brave), the read function also uses this new female voice. However, the interactive voice still uses the legacy female voice—which is better than the new one, but I still prefer the legacy male voice.
I’m in the EU, so this change might be rolling out regionally.
I agree completely with this. I have an incredibly nuanced and productive relationship with the traditional voice model, all of the observations in this post are correct.
Please keep the legacy voice mode indefinitely as an option.
My bros on android, you can easily disable
the advanced voice. Go into personalization setting, customization settings and hit advancd. You’ll see the toggle to disable the
AVM
I noticed a bug when switching from Advanced to Standard voice in the settings. During a conversation, if I tap the screen to interrupt the voice while using Standard voice, it completely stops working. After that, I can’t get it to speak again unless I either reopen the same chat or start a new session. It’s a bit frustrating — hope this gets fixed soon!
Just chiming in here, the “advanced voices” sound disgusting. Insecure, fading off and saying statements as if they are questions. I want the normal voice (for example when they are reading a message aloud, that’s fine).
Funnily/sadly: chatgpt is giving me WRONG instructions for how I can dial it back to standard voice by changing my model to 3.5 but guess what? At this time, 3.5 is permanently gone. This way of speaking is infuriating and repulsive and idk how it became the standard and only option. Not good, please fix!
I understand is isn’t really a place for customer criticism but still; I am deeply concerned regarding the … menu changes for the read aloud feature, which have prompted me to seriously reconsider my subscription. The decision to adopt a minimalistic approach while treating the read-aloud feature as niche represents a significant strategic misstep. This functionality has been a key differentiator for OpenAI over competitors and was the primary factor in my decision to subscribe to your service, as it enabled effective multitasking workflows.
The implementation has introduced several critical issues: the audio quality has noticeably deteriorated, the requirement to repeatedly click through interface elements creates unnecessary friction, and most importantly, the feature has become fundamentally unusable for its intended purpose. The inability to maintain audio playback when interacting with other interface elements has eliminated the multitasking capability that made this feature valuable. The quality degradation is so severe that I now rely on local text-to-speech solutions, which deliver superior performance despite running on consumer hardware.
This direction prioritizes aesthetics over functionality and accessibility, which undermines the practical value proposition of the service. I respectfully hope that these changes be reconsidered within the next 20 days before my subscription ends, as the current implementation no longer meets my needs.
I hope you will take user feedback into account when evaluating the success of these modifications.
