This project allows to use the full Power of AI. I used for this the Whisper model for speech to text recognition.
Then used the ChatGPT 3.5 - Turbo Model just to get the answers by streaming it directly. Then used the Bark model
to generate Audio from the answer. One problem i want to mention is that i ran this on my PC which for some reason
did not want to use the GPU, if it's you case then it will be very slow. So just make sure you run this on a GPU.
Don't hesitate to modify the code. For example the ask_stream function can be modified to display and launch audio for each word.
def ask_stream()
...
for response in openai.ChatCompletion.create(...):
#Code
text_to_audio(full_answer)
#Display word by word to get Typing effect
Can be changed to
def ask_stream()
...
for response in openai.ChatCompletion.create(...):
#Code
text_to_audio(answer_word)
#Display word