On Monday, OpenAI revealed that ChatGPT has undergone an update, introducing capabilities for voice conversations and image recognition. The company’s AI-powered chatbot will soon be able to understand photos users upload or share and deliver context or related information on all platforms. It will also be able to communicate using OpenAI’s Whisper voice recognition software and a new text-to-speech (TTS) engine that provides “human-like” audio on the ChatGPT mobile app.
The voice conversation feature will be accessible on iOS and Android via an opt-in setting, while OpenAI’s new picture recognition capability for ChatGPT will be accessible on all platforms. It remains unclear whether these features will be extended to users of the free tier, in addition to ChatGPT Plus and Enterprise members.
Select the Voice Conversations checkbox under Settings > New Features to enable ChatGPT voice conversations. Choose from five voices; OpenAI says it partnered with experienced voice actors for the new functionality. By translating your spoken inquiries into text, the chatbot can comprehend, the ChatGPT app can react, and the company’s new TTS technology will turn responses into audio.
On Monday, Spotify unveiled an AI-based speech translation tool for podcasters that automatically translates English to French, German, and Spanish. There will be more than just ChatGPT using OpenAI’s new TTS technology. Spotify said select podcasters are testing the tool, and translated episodes will be available to all users everywhere Spotify is.
The new image identification tool from OpenAI uses its GPT-3.5 and GPT-4 multimodal models to interpret images and text in photos, screenshots, and documents. Take a fresh image or share one from your phone to get ChatGPT insights.
The chatbot will also allow users to submit multiple photographs for debate, according to OpenAI. Mark an area of the image with the built-in drawing tool to focus it. If you draw a circle around an undone bicycle chain in a photo and send it to ChatGPT, it may be able to help.
