Important
This project has been retired and archived
If there is a need of continued use / development of this project for your own needs please feel free to fork the project - which will remain here in archived form.
This example connector allows you to establish a phone call with a Teneo bot. The connector makes your Teneo bot available via Twilio IVR (Interactive Voice Response). This guide will take you through the steps of setting a new Twilio phone number and deploying the connector to respond to events sent by Twilio.
When running the connector locally, making the connector available via https is preferred. Ngrok is recommended for this.
Your bot needs to be published and you need to know the engine url.
Below you will find two ways to run the this connector. The first way is by running the connector on Heroku. This is the easiest to get the connector running for non-developers since it does not require you to run node.js or download or modify any code.
The second way is to run the connector locally or to deploy it on a server of your choice. This preferred if you're familiar with node.js development and want to have a closer look at the code and plan to enhance or modify it.
Note: Starting November 28, 2022, free Heroku Dynos, free Heroku Postgres, and free Heroku Data for Redis® plans will no longer be available.
Click the button below to deploy the connector to Heroku:
In the 'Config Vars' section, add the following:
- TENEO_ENGINE_URL: The engine url of your bot
Optionally, you can also add the following parameters:
- LANGUAGE_STT: Speech to Text language. Defaults to en-US if not provided. For a list of supported languages, see: https://www.twilio.com/docs/voice/twiml/gather#languagetags. Note that your Teneo solution should match the chosen language as well.
- LANGUAGE_TTS: Text to Speech language. Defaults to en-US if not provided. For a list of supported languages, see: https://www.twilio.com/docs/voice/twiml/say/text-speech#amazon-polly. Note that your Teneo solution should match the chosen language as well.
Next, follow the Setup a Twilio phone number instructions. Make sure that in step 6 you set the Webhook to the url of the Heroku app.
If you want to run the connector locally, follow the steps below. If you have already followed the instructions above to deploy the connector on Heroku, you can skip this section and jump straight to Setup a Twilio phone number.
- Download or clone the connector source code:
git clone https://github.com/artificialsolutions/tie-api-example-twilio-ivr.git - Install dependencies by running the following command in the folder where you stored the source:
npm install - Create a
.envfile in the folder where you stored the source and add the URL of your engine. Optionally you can also specify the langauges for Speech To Text and Text To Speech:TENEO_ENGINE_URL=<your_engine_url> LANGUAGE_STT=en-US LANGUAGE_TTS=Polly.Joanna - Start the connector:
node server.js
Next, we need to make the connector available via https. We'll use ngrok for this.
- Start ngrok. The connector runs on port 1337 by default, so we need to start ngrok like this:
ngrok http 1337 - Running the command above will display a public https URL. Copy it, we will use it as a Webhook URL in step 6 of the Twilio Setup instructions below.
- Setup a free Twilio account, which comes with free credit money you can spend on buying a number.
- Go to the left pane menu and click the Programmable Voice phone icon, then from the submenu choose Numbers and click the Get a number button.
- Follow the instruction to obtain your phone number (the instruction may differ per region).
- Click on "Manage Numbers".
- Click your number so you can configure it.
- Under "Voice" -> "A call comes in" set the webhook to the Webhook URL you copied earlier.
That's it! Call your Twilio number with your phone, and speak to your bot!
The connector will send the following input parameters along with the user input to the Teneo engine:
The input parameter channel with value twilio is included in each request. This allows you to add channel specfic optimisations to your bot.
In case the user switched to dtmf input, the digits entered are provided in the input parameter digits.
The twilioConfidence input parameter contains the confidence score of the Speech To Text engine used by Twilio.
The input parameter twilioCallerCountry contains the country code of the caller. You can use this to improve the Speech to Text model, for example by adding an output parameter twilio_sttLanguage with value en-GB to responses for callers calling from the UK.
The connector will check for the following output parameters in an output to send specific data to Twillo:
If the output parameter twilio_customVocabulary exists, it will be used for custom vocabulary understanding or hints. The value should be a string. You may provide up to 500 words or phrases, separating each entry with a comma. Your hints may be up to 100 characters each, and you should separate each word in a phrase with a space. For more details: https://www.twilio.com/docs/voice/twiml/gather#hints
By default, end of speech is detected automatically. The twilio_customTimeout output parameter allows you to set a custom timeout. This can come in handy when you ask the user for a number for example. In that case you may prefer a longer end of speech detection timeout.
If the output parameter twilio_endCall with the value true exists, the call will be ended.
If the output parameter twilio_speechModel exists, it will be used to set a custom speech model. Allowed values are: 'default', 'numbers_and_commands' and 'phone_call'.
If provided, will override the environment variable LANGUAGE_STT.
If the output parameter twilio_inputType exists, it will be used to set a custom input type. Allowed values are: 'dtmf', 'speech' or 'dtmf speech'. DTMF allows the end user to enter a number using the keypad of the phone. User must press # to mark end of the input. The digits entered by the user will be sent to engine using an input parameter digits.