Create a conversational Voicebot on WhatsApp with Watson services

Conversational AI VoiceBots are helping businesses automate repetitive and time-consuming queries. This gives your team more time to tackle more complex, high priority strategic tasks, while still ensuring that the customers are engaged. Voicebots take in voice input can ingest, analyze, interpret, and respond to customer inquiries, with the end goal of either being able to answer a customer’s question in real-time or being able to continue a textual exchange with the customer to clarify the question at hand.

A Voicebot is useful in many scenarios where typing is not a convenient option:

• Elderly customers find it more convenient to speak than to type. Some customers have better spoken language skills than written language skills. They find it easier to use VoiceBots.
• Delivery executives on the move with a need to update systems.

In this code pattern, you will look a scenario of a customer speaking Brazilian Portuguese interact with a customer care center using WhatsApp. A framework is built using which the user will be able send queries using speech on WhatsApp, and then get a response from Watson Assistant. The query from the user is sent to the Watson Speech to Text service through a custom application. The output from the Watson Speech to Text service is then fed into Watson Assistant. Watson Assistant understands the query with the help of the built-in support for Brazilian Portuguese and responds back to the user with an appropriate reply.

When you have completed this code pattern, you will understand how to:

• Use Watson Speech to Text service.
• Use Watson Assistant service.
• Integrate WhatsApp with Watson services.
• Handle queries sent over voice on WhatsApp.

This code pattern can be configured to support multiple languages. Watson Assistant supports the following languages.

Flow

1. User sends a voice or text message in WhatsApp
2. The message is redirected to Twilio Programmable messaging service
3. Twilio redirects the message to the Voicebot Application deployed on IBM Cloud or OpenShift
4. If the user has sent a voice message, the Voicebot Application leverages Watson Speech to Text to transcribe the message into text
5. The text message is sent to Watson Assistant
6. Watson Assistant chatbot detects the intent and replies with response accordingly
7. The Voicebot Application triggers the Twilio Programmable messaging service with the response message
8. Twilio Programmable messaging service redirects the response message to WhatsApp
9. User will be able to view the response message on WhatsApp

Pre Requisites

1. IBM Cloud Account: Create an IBM Cloud account.
2. IBM Cloud CLI: Download and Install IBM Cloud CLI tool.
3. OpenShift ROKS Cluster: Create an OpenShift Cluster.
4. OC CLI: Download and Install OpenShift CLI tool.
5. Git Client: Download Git Client CLI.

Steps

1. Clone the repo
2. Create Watson Services
3. Build and Deploy the Voicebot Application
4. Create Twilio service
5. Configure Voicebot Application
6. Try out voicebot on WhatsApp

1. Clone the repo

Clone the /voicebot-on-whatsapp-using-watson-services repo locally. In a terminal, run:

git clone https://github.com/IBM//voicebot-on-whatsapp-using-watson-services.git

We will be using the following directory:

• voicebot-app/ - To build and deploy the application.

2. Create Watson Services

Create the following Watson Services:

• Watson Speech to Text
• Watson Assistant

2.1 Watson Speech to Text service

2.1.1 Create service

Watson Speech to Text service can be created on IBM Cloud Pak for Data as well as on IBM Cloud. Follow the steps accordingly to create service in either of the platform.

IBM Cloud Pak for Data

• On Cloud Pak for Data, go to the Service catalog and deploy Watson Speech To Text Service, by clicking on the Deploy and following the instructions.

IBM Cloud

• On IBM Cloud, create a Watson Speech To Text Service, under Select a pricing plan select Lite and click on create as shown.

2.1.2 Copy service credentials

• In Speech To Text Dashboard, Click on Services Credentials

• Click on New credential and add a service credential as shown.

NOTE: Once the credential is created, copy and save the credentials in a text file for using it in later steps in this code pattern.

2.2 Watson Assistant service

2.2.1 Create service

Watson Assistant service can be created on IBM Cloud Pak for Data as well as on IBM Cloud. Follow the steps accordingly to create service in either of the platform.

IBM Cloud Pak for Data

• On Cloud Pak for Data, go to the Service catalog and deploy Watson Assistant Service by clicking on the Deploy and following the instructions.

IBM Cloud

• On IBM Cloud, create a Watson Assistant Service, under Select a pricing plan select Lite and click on create as shown.

2.2.2 Copy credentials

• In Watson Assistant Dashboard, Click on Manage and Launch the Watson Assistant.

• Click on Create assistant and give your assistant a name and create it.

• In Assistant dashboard, under Dialog click on add dilog skill.

• Select Upload skill and upload the skill-Portugues-Assistant-Skill.json skill present in /watson-assistant-skill/ directory.

• In Assistant dashboard, click on the three dots and select Assistant settings.

• Copy the Assistant ID, Assistant URL and API key as shown.

NOTE: Copy Assistant ID, Assistant URL and API key and save the credentials in a text file for using it in later steps in this code pattern.

3. Build and Deploy the Voicebot Application

The Voicebot Application is an intermediator that connects Watson services and Twilio service. This application must be deployed in order to access it from Twilio WhatsApp Sandbox. You can either deploy the application on your OpenShift cluster or on any public hosting services like Cloud Foundry. Follow the documentation accordingly

Steps to Build and Deploy on OpenShift

Note: If you want to deploy the Voicebot Application without any modifications, you can skip the Build steps and directly follow the Deploy steps. If you want to make some modifications in the Voicebot Application and then deploy it then follow the Build step.

Build

Note: Make sure you have docker cli installed and logged in to DockerHub

• In cloned repo, navigate to voicebot-app/ directory and build the docker image. In terminal run:

$ docker build -t <your-docker-username>/voicebot-app:v1 .

Replace <your-docker-username> with your docker hub username

• Once the docker image is built, store the docker image to Dockerhub. In terminal run:

$ docker push <your-docker-username>/voicebot-app:v1

• At this point you have built the container image and successfully pushed to to a container repository dockerhub.

• Copy the image tag <your-docker-username>/voicebot-app:v1 and replace it on line no 18 in voicebot-app/deploy.yaml

spec:
      containers:
      - name: voicebot-app
        image: < your-docker-username >/voicebot-app:v1 
        ports:
        - containerPort: 8080

Deploy

• Login to your OpenShift cluster, In terminal run:

$ oc login -u <username> -p <password>

• Alternatively you can also login with an auth token. Follow the Step here to login through an auth token.

• Once you have logged into OpenShift from your terminal, you can run the oc apply command to deploy the application on OpenShift. From within the voicebot-app/ directory, in terminal run:

$ cd voicebot-app/
$ oc apply -f deploy.yaml

deployment.apps/voicebot-app created
service/voicebot-app-service created
route.route.openshift.io/voicebot-app-url created

• A deployment, service and a route will be created. To access the App, In terminal run:

$ oc get route -n default

NAME           HOST/PORT                                                                                                           PATH   SERVICES           PORT   TERMINATION   WILDCARD
voicebot-app-url   voicebot-app-url-default.xxx.us-south.containers.appdomain.cloud   /      voicebot-app-service   8080                 None

• You will see the PATH for with service name voicebot-app-url.
• Make a note of this url, as it will be used in step 4.

Example: http://voicebot-app-url-default.xxx.us-south.containers.appdomain.cloud

• At this point, you will have successfully deployed the voicebot app on OpenShift. Now lets access it and see how it looks like.

Steps to Build and Deploy on IBM Public Cloud Foundry

Build and Deploy

• Before you proceed, make sure you have installed IBM Cloud CLI in your deployment machine.

• From the cloned repo, goto voicebot-app directory in terminal, and run the following commands to deploy the Application to IBM Cloud Foundry.

  $ cd voicebot-app/

• Log in to your IBM Cloud account, and select an API endpoint.

  $ ibmcloud login

  NOTE: If you have a federated user ID, instead use the following command to log in with your single sign-on ID.

  $ ibmcloud login --sso

• Target a Cloud Foundry org and space:

  $ ibmcloud target --cf

• From within the voicebot-app directory push your app to IBM Cloud.

  $ ibmcloud cf push otp-api

• The manifest.yml file will be used here to deploy the application to IBM Cloud Foundry.

• On Successful deployment of the application you will see something similar on your terminal as shown.

Invoking 'cf push'...
Shown below is a sample output
    
Pushing from manifest to org [email protected] / space dev as [email protected]...
    
...
    
Waiting for app to start...
    
    name:              voicebot-app
    requested state:   started
    routes:            voicebot-app.xx-xx.mybluemix.net 
    last uploaded:     Sat 16 May 18:05:16 IST 2020
    stack:             cflinuxfs3
    buildpacks:        python
    
    type:            web
    instances:       1/1
    memory usage:    512M
    start command:   python app.py
        state     since                  cpu     memory           disk           details
    #0   running   2020-05-16T12:36:15Z   12.6%   116.5M of 512M   796.2M of 1

• Once the app is deployed, from the output of the above command, you can visit the routes to launch the application.

• Make a note of this url, as it will be used in step 4.

Example: http://voicebot-app.xx-xx.mybluemix.net if deployed on IBM Cloud Foundry, http://voicebot-app-url-default.xxx.us-south.containers.appdomain.cloud if deployed on OpenShift

Note: Add http:// to the OpenShift url if not present.

• At this point, you will have successfully deployed the Voicebot app on IBM Cloud. Now lets access it and see how it looks like.

4. Create Twilio service

Twilio is a SaaS offering that provides APIs to make and receive calls or text messages. As there are no APIs from WhatsApp directly availabe to send and receive WhatsApp messages programmatically, you will learn how to use Twilio's messaging service APIs that provides gateway to communicate with WhatsApp programmatically. Get started by creating a free Twilio service.

• Create a free Twilio service here: https://www.twilio.com/try-twilio.

NOTE: - Once you create a Twilio service, you will have to verify your email id as well as your phone number.

• You will receive verification link in the email provided during Twilio sign up. Go ahead and verify your email id.

• Once email id is verified you will be prompted to enter your phone number, submit that and you will get an OTP on your registered number, enter that back to verify.

  

• On successful verification you should see a welcome greeting message, additionally you will see some questions, select as described below.

  

  Questions	Answers
  Which Twilio product are you here to use?	WhatsApp
  What do you plan to build with Twilio?	IVR & Bots
  How do you want to build with Twilio?	With code
  What is your preferred coding language?	Python
  Would you like Twilio to host your code?	No, I want to use my own hosting service

• Visit the Whatsapp section in Twilio https://www.twilio.com/console/sms/whatsapp/sandbox

• You will see a popup box reqsuesting you to Activate Your Sandbox, click on I agree checkbox and click Confirm.

• The sandbox for WhatsApp will appear, make a note of the Sandbox Name which will be prefixed with join, click on Settings on the left panel and select WhatsApp Sandbox Settings.

• In WhatsApp Sandbox Settings page, under Sandbox Configuration, there will be a field called WHEN A MESSAGE COMES IN, replace the existing URL in that field with the URL obtained by deploying the Voicebot Application from Step 3, finally click on Save to save the configuration.

NOTE: Sometimes the changes are not saved in Twilio WhatsApp Sandbox Settings even after clicking on save, reload the page to enusre the URL that you have entered in WHEN A MESSAGE COMES IN field is reflecting over there. If you still see the old URL over there then enter the URL from Step 3 again and save it.

• Now the Voicebot Application is configured in Twilio, any message that you send from WhatsApp from this point will go to the Voicebot Application via Twilio WhatsApp Sandbox. However to reply back to you from WhatsApp the Voicebot Application needs to establish connection with Twilio.

• To establish connection between the Voicebot Application and Twilio we need to get the account_sid and auth_token from Twilio.

• Visit https://www.twilio.com/console and expand the Project Info tab. You will see the ACCOUNT ID and AUTH TOKEN, copy it in some notepad as it will be used in Step 5.

• At this point, you should have the Sandbox Name, account_sid and auth_token from Twilio service.

5. Configure Voicebot Application

Access the deployment URL created in previous step.

• You will see a configuration page as shown.

• You can see the Status of Twilio and Watson Services as Not Configured. Go ahead and click on Add Twilio Credentials. Enter the ACCOUNT SID and AUTH TOKEN copied in step 3.

• Once you submit, you will see the Status of Twilio service as Configured.

• Similarly click on Add Watson Assistant Credentials.Enter the APIKEY, URL and ASSISTANT-ID credentials copied in Step 2.

• Finally click on the Add Watson Speech to Text Credentials. Enter the APIKEY and URL copied in step 2.

• Once all the services are configured you will see a status as Configured.

• At this point you have successfully configured the Voicebot Application

6. Try out voicebot on WhatsApp

• Scan the QR code in your Phone to open the WhatsApp chat with Twilio's messaging API.

• A WhatsApp chat will open up in your phone with a typed code join <sandbox name>.

• Replace the <sandbox name> with your Sandbox Name obtained from Step 4 and send the message.

NOTE: If you are unable to scan the QR code, save the phone number +14155238886, open WhatsApp and send a message to the saved number with code join <sandbox name>.

You can follow the workflow of the app as follows:

• User can send Ola or Hi to start a Watson Assistant session.

• The user can either send a voice message or send a text message through WhatsApp's chat functionality.

• The Watson Assistant will respond with textual reply accordingly.

  

Summary

In this Code Pattern you learnt how to create a WhatsApp Chatbot that take in voice input and get a processed response from Watson Assistant. You also learnt how to use, Watson Assistant in conjunction with Watson Speech to Text.

This code pattern can be extended to any language.

Questions

If you have any questions or issues you can create a new issue here.

Pull requests are very welcome! Make sure your patches are well tested. Ideally create a topic branch for every separate change you make. For example:

1. Fork the repo
2. Create your feature branch (git checkout -b my-new-feature)
3. Commit your changes (git commit -am 'Added some feature')
4. Push to the branch (git push origin my-new-feature)
5. Create new Pull Request

License

This code pattern is licensed under the Apache License, Version 2. Separate third-party code objects invoked within this code pattern are licensed by their respective providers pursuant to their own separate licenses. Contributions are subject to the Developer Certificate of Origin, Version 1.1 and the Apache License, Version 2.

Apache License FAQ