The app can do speech-to-text, and text-to-speech using their whisper model, and it's integrated with chatgpt. On May 13th, they demoed GPT-4o's multi-modal vo

The app can do speech-to-text, and text-to-speech using their whisper model, and it's integrated with chatgpt. On May 13th, they demoed GPT-4o's multi-modal voice capabilities, and it's much better than a text/speech model. Converting speech to text, and vice versa, is quite lossy, all the tone and emotion of the speech is lost in translation, the model they demoed in May was capable of much more than whisper is.