A new project called VoiceGPT has been launched, which is a voice assistant that utilizes the powerful ChatGPT chatbot to provide realistic, synthesized speech responses to voice commands. The system works by collecting audio of the speaker's voice using a USB microphone, which is then converted into text using the Google Cloud Speech-to-Text API. The text is then queried against ChatGPT using an unofficial API, which returns a text string of ChatGPT's response. The response is then processed by Google Cloud's Text-to-Speech API to turn it into realistic, synthetic speech that the Raspberry Pi can play through a speaker.
The project was developed by Nick Bild, who chose the Raspberry Pi 4 single board computer to host the project due to its versatility and Linux operating system. While the concept of a voice assistant is not new, ChatGPT provides a far better experience than anything currently on the market. In the future, the creator plans to add a keyword spotting algorithm to the project so that it can run in the background, waiting for a specific keyword to wake up.
Bild acknowledges that before he has the chance to add much more to the project, there will likely be a commercial product that includes ChatGPT on the backend. However, he remains enthusiastic about the possibilities of the project and is excited to see what others will create with it. The VoiceGPT project is available on GitHub for anyone interested in exploring its potential.
The project can be easily built by anyone who has a Raspberry pi, a USB microphone and a speaker. Complete details of the project have been uploaded on the GitHub repository which you can check out. Keep tinkering.