ChatGPT announced their new voice model: GPT Bidi 1!

Many of us, when we want to record professional audio, either have to go to a studio or record at home with a high-quality microphone. However, in both scenarios, we have to pay very serious amounts of money.



Nowadays, this problem is being solved thanks to artificial intelligence. First ElevanLabs, then Suno and Gemini, and now GPT is joining this race. OpenAI recently introduced GPT Bidi 1, a new voice model coming soon to ChatGPT. With this move, GPT has shown that it is moving beyond being just an AI you use for messaging and is trying to be actively present in every aspect of your life.

While GPT Bidi 1 doesn't currently produce voiceovers like ElevenLabs or music like Suno, we understand that OpenAI has similar intentions with the Bidi 1 model. So what is Bidi 1?

GPT Bidi 1, or "bidirectional" for short, allows you to communicate in two directions. Existing voice assistants would often interrupt us when we stopped talking, or they'd be at a loss for words. Bidi 1, however, has human-like reflexes, as if you were speaking to a human. If you interrupt while speaking, it will respond with "okay" or "uh-huh," letting you know it's listening.



From our experience, it's incredibly flexible and has human-like reflexes. The way it indicates it's listening while you're speaking, and how it responds to requests for something else immediately after answering, is quite surprising.

Furthermore, one of the best aspects of this model is the elimination of memory issues. Existing models tend to forget previous answers and the beginning of conversations as the conversation gets longer. With the GPT Bidi 1 model, this problem is eliminated.