Uses OpenAI GPT2 with a language model head provided by hugging face transformers to predict human-like response given a human speech input. Initial speech to text uses Google's Speech Recognition API while the text to speech conversion makes use of Google's Text To Speech API.
python == v3.9
To install packages, git clone the repo. Open command prompt and navigate to the root dir of where the repo is stored. In the command prompt type:pip install -r requirements.txt
Open command prompt and navigate to where the repo is stored and type:python main.py
There are options to select different model sizes
small: python main.py --s small
medium: python main.py --s medium
large: python main.py --s large
Greet Bot by saying: 'Hey Kai'
To exit loop press 'esc' key, else it will leave the loop after 10 input-response loops
Best used in places with lower background noise