OpenAI is offering free ChatGPT users a sneak peek of its impressive Advanced Voice mode This exclusive preview, confirmed at the company's Dev Day event in San Francisco, gives non-subscribers a brief overview of the differences between basic voice and
What sets Advanced Voice apart from the free version and Google's new Gemini Live is that it is native speech synthesis This means that instead of converting what you say into text, analyzing that text, and sending it back to you as speech, it simply listens to what you say and understands the nuances of your tone of voice and even emotional traits
I suspect that at some point in the future, if OpenAI succeeds in improving the underlying model and making it less expensive, they will roll out Advanced Voice more fully to all users
If you subscribe to ChatGPT Plus, which costs $20/month, there is a price limit on Advanced Voice usage However, I have been using it quite extensively for over a month and have yet to reach the limit
It is difficult to explain why Advanced Voice is so much better than Gemini Live or Meta's new AI Voice until I actually use it Gemini Live is very impressive, and Google's engineers have succeeded in capturing natural voice and allowing it to be interrupted in real time, but it lacks something special
For example, when I was showing Advanced Voice to my 3-year-old son, I said, “Hi, ChatGPT, this is my son, he's 3 years old, his name is Theodore,” and the tone of his voice immediately changed to what it sounds like when talking to a small child He spoke directly to my son and I knew if it was me or my son
Another impressive feature is the ability to change the accent and lock that accent into memory so that it continues to speak in that accent every time This includes making them speak like Yoda, the pirate, or Yoda's pirate!
Also, the potential of native speech synthesis is rarely seen due to safety concerns and the guardrails set up in the underlying model Very occasionally, we get a glimpse of its potential when it starts beatboxing or singing In the future, as OpenAI discovers ways to make safety less of an issue, we will see more features like this
All that said, if you have the free version of ChatGPT and get the Advanced Voice sneak peek, I encourage you to try it
Comments