Transferred fromBso89comAI Cambrian
To achieve telephone calls, you need at least the following key capabilities: speech recognition; speech synthesis; 3, real-time interaction; all these technologies are available in OpenAI, and now it depends on the final product.
OpenAI seems to be trying to make phone calls in chatGPT. This may be a small part of the event announced on Monday, as can be found in some back-end code on the OpenAI website
Now, OpenAI is also equipped with webRTC servers to achieve this goal, and these servers have also been configured recently
At first, people thought that webrtc was used in their pure voice mode, but it turned out to be done by Livekit.
As early as February, the former head of developer relations at OpenAI mentioned that the ultimate form of ChatGPT is not chat, but seems to correspond to the current information.
To make a phone call, you need at least the following key capabilities:
1. Speech recognition-convert speech signal to text
2. Speech synthesis-convert text to natural speech output
3. Real-time interaction-the ability to process and respond to each other's voice input in real time during a call
OpenAI has all these technologies, and now it depends on the final product.
The head of OpenAI Voice Technology also retweeted tweets from Monday's press conference.
