AI - creating a voiced role-play conversational output from ChatGTP

Introduction

Code

the following is derived from https://www.youtube.com/watch?v=5htSGLCroc8

create an index.html file under a folder called templates to allow user inputs:

<!DOCTYPE html>
<html>
<head>
   <title>ChatBot</title>
</head>
<body>
  <h1>AI Director ED</h1>
  <form method="POST" action="/send_message">
     <input type="text" name="human_input" placeholder="Enter your message">
     <button type="submit"> Send </button>
     </form>
  <div id="response_message"></div> <!--this displays the response from ChatGPT -->
  
  <script>
     const form = document.querySelector('form');
     const response_message = document.get_ElementById('response_message');
     
     form.addEventListener('submit', function(e) {
        e.preventDefault();
        const formData = new FormData(form);
        fetch('/send_message', {
           method:POST,
           body: FormData
           }
        };
        .then(response => response.text() )
        .then(data => {
           responseMessage.innerHTML = data;
           }
           );
           
        form.reset();
        }
        );
   </script>
  </body>
</html>

python code set up

from langchain import OpenAI, LLMChain, PromptTemplate
from langchain.memory import ConversationBufferWindowMemory #as we need to store history
from dotenv import find_dotenv, load_dotenv #load your API_Keys for OpenAI and ElevenLabs which you stored in env.py
import requests #for http requests to OpenAI
from playsound import playsound #play and audio file
import os

load_dotenv(find_dotenv)

text to speech API

in this case we are using one from ElevenLabs - need to sign up and get an API_KEY but there is a free access level which allows use of APIs

ELEVEN_LABS_API_KEY = os.getenv("ELEVEN_LABS_API_KEY")  #this is the one you have stored in env.py file under that element name

def get_voice_message(message):
    payload = {
       "text" = message,
       "model_id" = "eleven_monolingual_v1",
       "voice_settings" = {
          "stability" = 0,
          "similarity_boost" = 0
       }
    }     
       
    headers = {
      'accept': 'audio/mpeg',
      'xi_api_key':  ELEVEN_LABS_API_KEY,
      'Content_Type': 'application/json'
    }
    response = requests.post('the URL from ElevenLabs API for your selected voicing', json=payload, headers=headers)
    if response.status.code == 200 and response.content:
       with open('audio.mp3', 'wb') as f:
         f.write(response.content)
       playsound('audio.mp3')
       return response.content

get human input into ChatGPT

# work out what context prompt you want to send to ChatGPT to set up a scenario  for the converstion - see flowgpt.com for assistance

#define a function to respond to human input:

def get_response_from_AI(human_input):
    template = '''
    you are as a role of a manager of an emergency department in a hospital, now lets play with the following requirements:
    1/ your name is Gary, you are 62 years old, you have worked in the hospital for 30 years and are now planning to retire soon as you like camping
    2/ the emergency department is an extremely busy and stressful environment
    3/ don't be cruel, don't be aggressive, don't be overly negative, don't be too boring, be enthusiastic and dynamic; 
    
    {history}
    Employee: {human_input}
    Gary:    
    
    '''
    
    prompt = PromptTemplate(
             input_variables=("history","human_input"),
             template=template
             )
    chatgpt_chain = LLMChain(
                    llm=OpenAI(temperature=0.2),
                    prompt=prompt,
                    verbose=True,
                    memory=ConversationBufferWindowMemory(k=2)
                    )
    output = chatgpt_chain.predict(human_input=human_input)
    
    return output

create web GUI using the index.html file

from flask import Flask, render_template, request

app = Flask(__name__)

@app.route("/")
def home():
    return render_template("index.html")
    
@app.route('/send_message',methods=['POST'])
def send_message():
    human_input = request.form('human_input')
    message=get_response_from_AI(human_input)
    get_voice_message(message) #get and play the text to speech audio
    return message
    
if __name__ == "__main__":
   app.run(debug=True)
   
   
#if all this python code in the cells above are saved to a file called app.py then it can be run via python app.py

OzEMedicine - Wiki for Australian Emergency Medicine Doctors

Table of Contents

AI - creating a voiced role-play conversational output from ChatGTP

Introduction

Code

create an index.html file under a folder called templates to allow user inputs:

python code set up

text to speech API

get human input into ChatGPT

create web GUI using the index.html file