I want to add serious feedback about the current ChatGPT Voice/Live experience

I want to add serious feedback about the current ChatGPT Voice/Live experience. The problem is not simply that I liked an older voice better. The previous Standard Voice / integrated spoken response experience worked as a cognitive interface. It allowed me to speak or write in the same chat and receive spoken responses that preserved the assistant’s reasoning, memory, custom instructions, tone, personality, and continuity. The current Voice/Live experience feels like a separate layer inserted into the conversation. When I switch from a serious written chat into Voice, the assistant can become less accurate, less contextual, less personalized, and less aligned with prior instructions. This breaks the workflow and creates a trust problem. The interface makes it look like the same assistant is continuing the same conversation, but the behavior suggests that Voice may have different effective context handling, reasoning depth, tool access, or instruction-following. That should be clearly disclosed. For professional users, accessibility-dependent users, researchers, writers, strategists, and people using ChatGPT as a long-form thinking partner, this is not a minor UX change. Voice is part of the cognitive workflow. Please restore a true Classic / Standard Voice option that speaks the normal written ChatGPT response with the same reasoning, memory, custom instructions, personality, and transcript continuity, without forcing users into Live Mode.