If you pay for ChatGPT, you’ll soon be able to do more than just communicate with the bot over text. In the next two weeks, paid subscribers will get access to new features including image and voice prompts. Non-paying users will have to wait longer, although parent company OpenAI does plan to spread the fun to everyone eventually.
You’ll be able to speak your query to ChatGPT and it will turn your speech into text using OpenAI’s Whisper model to feed into the large language model. Then it will convert the answer it receives from text into speech to read it to you. There will be five different voice options to choose from.
The image prompts will allow you to send the bot an image and ask a question about it (an example from OpenAI shows a user sending a photo of their bicycle and asking how to lower the seat). You’ll be able to draw on the image if necessary to further clarify what you want to know. For accuracy and privacy reasons, ChatGPT will not be able to “analyze and make direct statements about people” using this feature.