ChatGPT’s new voice integration feels like the missing piece in AI chat (www.techradar.com)

OpenAI quietly rolled out a major UX upgrade: ChatGPT Voice is now fully integrated into the regular chat interface on mobile and web, so you no longer leave your conversation to enter a separate “Voice Mode.” Tap to speak, get live transcriptions in the chat, interrupt or switch back to typing at any time, and summon visuals like maps, weather tables, news links, or camera-based answers without breaking context. Voice-driven actions can update saved memory (e.g., “remember I live in…”), and you can even ask for image generation by voice — though reviewers report occasional failures on that feature. If you prefer the old floating-orb experience, a “Separate mode” toggle remains in Settings. This matters because it removes friction that previously made voice feel like a special case, turning it into a seamless, multimodal interaction loop that’s more responsive than traditional assistants (Alexa/Siri) and competitive with Gemini Live. Technically, the update emphasizes persistent conversational context, interruptible streaming responses, real-time transcript rendering, and inline multimodal outputs — all of which enable hands-free, context-aware workflows and better accessibility. Early reliability kinks (image generation delays) and latency/robustness will be important to watch, but the change signals a shift toward voice-as-default in AI interfaces and tighter convergence of text, audio, vision, and live web content.