ChatBot with Open AI, LangChain and PlayHT

Main Handbook, This Handbook

Prerequisites

• Existing Open AI Account

  • Check whether you have a limit to using the OpenAI API

  • The below screenshot indicated it is expired by June 1. So, you need to create a new account for free credits or purchase premium plan.

    

• Creating OpenAI Account with New Mail or Mobile Number

  • Open https://openai.com

  • Click on Sign Up button

  • Choose your preffered Sign Up method

  • After logged in click here to see below options

    

• Create a HuggingFace Account

  • Open https://huggingface.co/

  • Click on Sign Up button

  • Enter your details and Click on Sign Up

  • Verify your email address

• Create a New Space in HuggingFace Account

  • Click on Profile icon top right corner

    

  • Click on New Space

    

  • Enter details of your new space

    

• Create PlayHT Account

  • Open https://play.ht/

  • Click on Log in

  • Since you don't have an account, go to https://play.ht/signup/

  • Choose your preferred Sign Up method

  • Account will be created and you will be redirected to your dashboard

• Clone your Voice to PlayHT

  • Before cloning, get ready with a 45 second clean audio of your voice

  • Open https://play.ht/studio/voice-cloning/

  • Click on + Create a New Clone

  • Click on Instant

  • Enter Voice Name and Upload the File

  • Click on Create

  • Your voice will be cloned

• Open the below provided Colab link

  

• Copying Code to your Google Drive

  • On the top left corner of Google Colab Notebook you can find File, click on it

    

  • Click on Save a copy in Drive

    

  • If you are not logged in to your Google Account, please log into it

  • Once you are successfully logged in a new Google Colab Notebook with the given code will be opened

Install the Required Packages

• Click on Play button to Install the Packages

  

Import the Required Packages to build the application

• Click on Play button to Import the Packages

  

Assign API Keys and User IDs

• How to get OpenAI API Key?

  • Go to https://platform.openai.com/account/api-keys

  • Click on + Create new secret key

  • Copy the secret key for your use

• Assign OpenAI API Key

  • Replace OPENAI_API_KEY with your OpenAI API Key

    

• How to get PlayHT User ID and PlayHT API Key?

  • Login to PlayHT - https://play.ht/

  • Go to https://play.ht/studio/api-access/

  • Generate the Secret Key

    

  • Replace PLAY_HT_API_KEY with Secret Key you copied

    

  • Replace PLAY_HT_USER_ID with User ID you copied

    

Set OpenAI API Key to Environmental Variables

• Click on Play button to set it as a Key

  

Assign your Voice to a variable and set the URL to a variable

• How to get your Voice ID

  • Go to https://docs.play.ht/reference/api-list-cloned-voices

  • In the right side section, select the language Python

    

  • Enter Authorization as Bearer PLAY_HT_API_KEY replace PLAY_HT_API_KEY with your API Key refer here

    

  • Enter X-USER-ID as your PLAY_HT_USER_ID refer here

    

  • Click on Try it!

    

  • If the Response says 200 then you have successfully created your Voice ID

    

  • Copy the id from the Response like s3://.../.json

  • Now you can use that id in the get_payload as a value for voice

  • Replace PLAY_HT_VOICE_ID with your copied id

    

  • Click on Play button to set the values

    

Assigning the values for template, prompt, and memory

• You can update the first line of the template You are a helpful assistant to answer user queries with <Prompts> given in Prompts and Examples

• Click on Play button

  

Initializing LLM Chain using OpenAI

• Using ChatOpenAI method we are creating an LLM Chain

• Click on Play button

  

Setting headers

• Click on Play button

  

Define functions to fetch Audio Response from PlayHT

• Click on Play button

  

Define a function to generate the response for the question you ask

• Click on Play button

  

Define a function to generate the Audio response and Create a ChatInterface using the Gradio

• We are defining a chat_bot_response to return either an audio response format if audio file is fetched properly or returns a string response if it is an error

• We are creating the ChatInterface from gradio and providing a function chat_bot_response and also examples

• Check for other arguments here

• Click on Play button to create an interface

  

Launch your ChatBot with Gradio App

• Click on Play button to launch the App

  

Now you can try asking questions in your ChatBot and receive response as your Voice

If you are getting any errors

• Keep print statements to identify the issue

• To identify the error you are getting please add debug = True while launching the gradio app

  if __name__ == "__main__":
      demo.launch(debug = True)

• You can add try and except block to handle the errors

  def get_text_response(user_message,history):
      try:
          response = llm_chain.predict(user_message = user_message)
      except Exception as e:
          print("Error:", e)
          try:
              print("Error:", e.error.message)
              response = "Failed to reply: " + e.error.message
          except Exception as e:
              response = "Failed to reply"
      return response

Publish your code to Hugging Face

Login to Hugging Face from Google Colab

• Create a Hugging Face token and Copy

  • Login to Hugging Face https://huggingface.co/

  • Open https://huggingface.co/settings/tokens

  • Click on New token

  • Add a Name for the Token

  • Choose write Role for the Token

  • Click on Generate a token

  • Copy the Token

• Click on Play button to enter Hugging Face Token

  

• Now paste the Hugging Face token in the textbox provided and click on Login

  

Create Hugging Face API to push code from Google Colab

• Click on Play button to create API

  

Adding Hugging Face Repository ID

• Copy Hugging Face Repository ID by opening the Hugging Face Repo Created

  

• Replace your Repo ID

  

• Click on Play button to assign Hugging Face Repo ID

  

Adding APIs to Hugging Face secrets

• Click on Settings Button

  

• Go to Variables and secrets section

  

• Click on New secret

  

• Enter Name as OPENAI_API_KEY and Value as your OpenAI API Key

  

• By the similar process create secrets for PLAY_HT_API_KEY, PLAY_HT_USER_ID and PLAY_HT_VOICE_ID

Load App and Requirements file

• Click on Play button to download files

  

• You can check downloaded files in the left side panel

  

• You can edit the downloaded files

  • Click on the app.py file

    

  • The file will be opened and you can edit the template section by referring to Prompts and Examples

    

  • After editing the file don't forget to save, by pressing ⌃ + S or ⌘ + S

Push your code to Hugging Face

• Click on Play button

  

• Now in your space, you should see it is Building status

  

• On Succesful Build, you should see Running status

  

• If the Build fails you will see a Runtime error with some errors below

  

• Click on Logs to check more details about the error

---

Show some ❤️ by starring ⭐ this repository!