OpenAI has expanded ChatGPT’s voice mode with two features: anyone talking to the AI can now upload a file and ask questions about it, and voice can for the first time be used inside Projects too. The full feature is also new for Business, Enterprise, and Edu accounts, which previously had no voice access at all.
ChatGPT now listens to uploaded documents too
Until now, ChatGPT’s voice mode worked like a phone call: you talked, the AI answered, and documents stayed out of the picture. According to OpenAI’s release notes, you can now upload a file mid-conversation – a PDF, a Word document, or a photo of a printout – and the AI reads the content and answers questions about it.
Voice mode can also now be used inside “Projects” – the topic folders where ChatGPT bundles several chats, files, and fixed instructions for a task. Someone who has set up a project folder for monthly billing, for instance, can access the documents stored there by voice instead of explaining the context every single time.
The voice is powered by the GPT-Live-1 model, which OpenAI says can listen and speak at the same time. That makes interruptions and follow-up questions feel more natural than with the previous voice feature.
On the go, voice mode handles document checks
An accounting clerk gets a PDF invoice by email on the way to an appointment. Instead of stopping to type, she opens ChatGPT, switches on voice mode, uploads the invoice, and asks: “What’s the total including VAT, and does the IBAN match the one from last time?” The AI reads the document and answers straight away by voice – the check is done before her train leaves.
Here’s how to get started:
- Open the ChatGPT app on your phone, in the browser, or on desktop.
- Turn on voice mode using the microphone icon.
- During the conversation, tap the upload icon and pick the file you need.
- Ask your question about the file out loud, for example about totals, deadlines, or discrepancies.
- Keep working in the linked project folder afterward if needed.
The same trick works for meeting minutes, contracts, or quotes: anyone on the move or with both hands full gets the content read aloud and can ask follow-up questions right away.
Business accounts now get the feature too
Until now, voice mode was mostly reserved for personal accounts – Business, Enterprise, and Edu accounts were left out at launch in July. That has changed: according to OpenAI, the feature, including file upload and Project access, is now rolling out worldwide to business and education accounts as well.
Free accounts, for which OpenAI has lifted the message limit since August 10, use the simpler GPT-Live-1 mini model, while paid accounts – Plus, Pro, Business, and Enterprise – get the fuller GPT-Live-1 version. There’s no extra charge for the feature itself. Business accounts typically need IT to switch on access, similar to how it already works with the new ChatGPT slides for PowerPoint and its business credits.
Voice mode is available in the ChatGPT app for Mac, Windows, iOS, and Android, and in the browser at chatgpt.com. The expanded voice feature could previously only be reached in Europe via workarounds – that hurdle has since been removed.
What matters now is how reliably the AI holds up with longer or messier documents – earlier voice versions struggled with complex tables and fine print. Anyone having customer or contract documents evaluated by voice should first check with IT or compliance whether internal data rules allow it. GPT-Live’s original launch already sparked discussions about the psychological risks of humanized AI voices – those concerns remain with the new features. It’s also still unclear when OpenAI will name a firm start date for Germany – so far the company has only mentioned a worldwide rollout without a calendar date.
