I tested ChatGPT's Live Voice upgrade, and it almost felt human - how to try it
ChatGPT's new live voice mode can listen, speak, and browse the web simultaneously, creating a near-human interaction experience.
- ChatGPT's Live Voice now combines real-time listening, speaking, and web browsing for seamless human-like interactions.
- The upgrade reduces latency and enables the AI to answer questions about current events without prior training.
- Available exclusively to ChatGPT Plus subscribers with compatible hardware.
- Early feedback highlights improved tone modulation and reduced unnatural pauses.
OpenAI's latest ChatGPT Live Voice upgrade introduces a multimodal capability that allows the AI to process speech inputs, generate natural-sounding responses, and perform live web searches all in real time. This creates an interaction that closely resembles a human conversation, with the AI adapting dynamically to user queries without noticeable delays.
The upgrade builds on previous voice features by integrating real-time information retrieval, enabling the model to answer questions about current events or niche topics without prior training on those subjects. Users can now engage in back-and-forth dialogue while the system simultaneously fetches relevant data from the internet, reducing the risk of outdated or incorrect responses.
Early testers report that the latency between speech and response is minimal, and the AI's tone modulation feels more natural than earlier versions. The feature is currently available to ChatGPT Plus subscribers and requires a compatible device with a microphone and internet connection.
Demonstrates advancements in real-time multimodal AI interactions, useful for building conversational agents.
Offers a competitive edge for customer service and interactive applications requiring natural dialogue.
Showcases cutting-edge AI capabilities in human-computer interaction.
Bridges the gap between AI and human conversation, making interactions feel more natural.
- multimodal AI
- AI systems that process and generate multiple types of data, such as text, speech, and images, simultaneously.
Changing the AI conversation in adult education - Auburn University
AI ToolsMeta enters the crowded AI coding battle with Muse Spark 1.1
AI ToolsDidit Copilot: shipping an AI agent that lives inside the app
AI ToolsOpenAI pairs its GPT-5.6 public rollout with ChatGPT Work, a new agent that handles entire workflows
AI ToolsEnhancing enterprise inference on Amazon SageMaker HyperPod with data capture, Hugging Face, NVMe, and Route 53 integration
SecurityNew York Times says OpenAI hid evidence in ChatGPT copyright trial
The New York Times has filed a motion alleging OpenAI concealed tools and datasets that could reveal copyrighted journalism in ChatGPT's outputs, intensifying their legal dispute.
The AI Republic? Governing Artificial Intelligence at America’s 250th Anniversary - Foley & Lardner LLP
Foley & Lardner LLP explores the governance of artificial intelligence in the United States as it approaches its 250th anniversary.
SecurityOpenAI may have made a fatal misstep in copyright fight with news orgs
OpenAI is accused of deleting or hiding ChatGPT interaction logs during its copyright dispute with the New York Times, potentially violating legal holds.
AI system developed by UC Irvine physicists helps explain why neutrinos have mass - UC Irvine News
Researchers at UC Irvine have developed an AI system that helps explain why neutrinos have mass. The breakthrough was made possible by the AI's ability to analyze complex data.
BusinessGoogle will now disclose which ads are made with AI
Google will start labeling ads created or edited with generative AI, aiming to improve transparency for users and advertisers.
LLMGPT-5.6 Sol nearly matches Fable 5 on aggregated benchmarks at one-third the cost
OpenAI's GPT-5.6 Sol scores 59 on the Artificial Analysis Intelligence Index, just one point behind Fable 5, while costing a third of the price.