Get started with Voice on Android:
- Quickstart - Run the quickstart app
- Emulator Support - Android emulator support
- Reducing APK Size - Use ABI splits to reduce APK size
- Access Tokens - Using access tokens
- More Documentation - More documentation related to the Voice Android SDK
- Twilio Helper Libraries - TwiML quickstarts.
- Issues & Support - Filing issues and general support
To get started with the Quickstart application follow these steps. Steps 1-6 will allow you to make a call. The remaining steps 7-8 will enable push notifications using FCM.
- Open this project in Android Studio
- Create a Voice API key
- Configure a server to generate an access token to use in the app
- Create a TwiML application
- Configure your application server
- Run the app
- Generate google-services.json
- Add a Push Credential using your FCM Server API Key
- Receiving an Incoming Notification
- Make client to client call
- Make client to PSTN call
Go to the API Keys page and create a new API key.
Save the generated API_KEY and API_KEY_SECRET in your notepad. You will need them in the next step.
Download one of the starter projects for the server.
- voice-quickstart-server-java
- voice-quickstart-server-node
- voice-quickstart-server-php
- voice-quickstart-server-python
Follow the instructions in the server's README to get the application server up and running locally and accessible via the public Internet. For now just add the Twilio Account SID that you can obtain from the console, and the API_KEY and API_SECRET you obtained in the previous step.
ACCOUNT_SID = 'AC***'
API_KEY = 'SK***'
API_KEY_SECRET = '***'
Next, we need to create a TwiML application. A TwiML application identifies a public URL for retrieving TwiML call control instructions. When your Android app makes a call to the Twilio cloud, Twilio will make a webhook request to this URL, your application server will respond with generated TwiML, and Twilio will execute the instructions you’ve provided.
To create a TwiML application, go to the TwiML app page. Create a new TwiML application, and use the public URL of your application server’s /makeCall endpoint as the Voice Request URL.
As you can see we’ve used our ngrok public address in the Request URL field above.
Save your TwiML Application configuration, and grab the TwiML Application SID (a long identifier beginning with the characters "AP").
Put the remaining APP_SID configuration info into your application server by opening server.py and setting the following constants with the information you gathered above.
ACCOUNT_SID = 'AC***'
API_KEY = 'SK***'
API_KEY_SECRET = '***'
APP_SID = 'AP***'
Once you’ve done that, restart the server so it uses the new configuration info. Now it's time to test.
Open up a browser and visit the URL for your application server's Access Token endpoint: https://{YOUR_SERVER_URL}/accessToken. If everything is configured correctly, you should see a long string of letters and numbers, which is a Twilio Access Token. Your Android app will use a token like this to connect to Twilio.
Paste the public URL of your application server’s https://{YOUR_SERVER_URL}/accessToken endpoint into TWILIO_ACCESS_TOKEN_SERVER_URL in VoiceActivity.java. Make sure to include /accessToken in the URL path.
Run the quickstart app on an Android device
"
Press the call button to open the call dialog.
Leave the dialog text field empty and press the call button to start a call. You will hear the congratulatory message. Support for dialing another client or number is described in steps 10 and 11.
The Programmable Voice Android SDK uses Firebase Cloud Messaging push notifications to let your application know when it is receiving an incoming call. If you want your users to receive incoming calls, you’ll need to enable FCM in your application.
Follow the steps under Use the Firebase Assistant in the Firebase Developers Guide. Once you connect and sync to Firebase successfully, you will be able to download the google-services.json for your application.
Login to Firebase console and make a note of generated Server API Key and Sender ID in your notepad. You will need them in the next step.
"
Make sure the generated google-services.json is downloaded to the app directory of the quickstart project to replace the existing app/google-services.json stub json file. If you are using the Firebase plugin make sure to remove the stub google-services.json file first.
As a final step re-run the application from Android Studio to ensure the APK now has the latest google-services.json file.
You will need to store the FCM Server API Key with Twilio so that we can send push notifications to your app on your behalf. Once you store the API Key with Twilio, it will get assigned a Push Credential SID so that you can later specify which key we should use to send push notifications.
Go to the Push Credentials page and create a new Push Credential.
Paste in the Server API Key and press Save.
"
Put the PUSH_CREDENTIAL_SID configuration info into your application server by opening server.py and setting the following constants with the information you gathered above.
ACCOUNT_SID = 'AC***'
API_KEY = 'SK***'
API_KEY_SECRET = '***'
PUSH_CREDENTIAL_SID = 'CR***'
APP_SID = 'AP***'
Once you’ve done that, restart the server so it uses the new configuration info. Now it's time to test. Use your browser to initiate an incoming call by navigating to the public URL of your application server’s https://{YOUR_SERVER_URL}/placeCall endpoint. This will trigger a Twilio REST API request that will make an inbound call to your mobile app. You will receive an incoming call notification.
"
If your applicaion is in the foreground, you will see an alert dialog.
"
Once your app accepts the call, you should hear a congratulatory message.
To make client to client calls, you need the application running on two devices. To run the application on an additional device, make sure you use a different identity in your access token when registering the new device. Press the call button to open the call dialog.
Enter the client identity of the newly registered device to initiate a client to client call from the first device.
A verified phone number is one that you can use as your Caller ID when making outbound calls with Twilio. This number has not been ported into Twilio and you do not pay Twilio for this phone number.
To make client to number calls, first get a valid Twilio number to your account via https://www.twilio.com/console/phone-numbers/verified. Update server.py and replace CALLER_NUMBER with the verified number. Restart the server so it uses the new value.
Press the call button to open the call dialog.
Enter a PSTN number and press the call button to place a call.
The SDK supports using emulators except in the following known cases:
- Emulators with API 22 or lower have bad audio emulation, the sound is generally inaudible
- Emulators must have Google Play services support to use FCM to receive call invites
- Running on x86 API 25 emulators results in application crashes
In general we advise using a real device when doing development with our SDK since real-time audio is a performance oriented operation.
Our library is built using native libraries. As a result, if you use the default gradle build you will generate an APK with all four architectures(armeabi-v7a, arm64-v8a, x86, x86_64) in your APK.
APK splits allow developers to build multiple APKs for different screen sizes and ABIs. Enabling APK splits ensures that the minimum amount of files required to support a particular device are packaged into an APK.
The following snippet shows an example build.gradle with APK splits enabled.
apply plugin: 'com.android.application'
android {
// Specify that we want to split up the APK based on ABI
splits {
abi {
// Enable ABI split
enable true
// Clear list of ABIs
reset()
// Specify each architecture currently supported by the Voice SDK
include "armeabi-v7a", "arm64-v8a", "x86", "x86_64"
// Specify that we do not want an additional universal SDK
universalApk false
}
}
}
The adoption of APK splits requires developers to submit multiple APKs to the Play Store. Refer to Google’s documentation for how to support this in your application.
The access token generated by your server component is a jwt that contains a grant for Programmable Voice, an identity that you specify, and a time-to-live that sets the lifetime of the generated access token. The default time-to-live is 1 hour and is configurable up to 24 hours using the Twilio helper libraries.
In the Android SDK the access token is used for the following:
- To make an outgoing call via
Voice.call(Context context, String accessToken, String twiMLParams, Call.Listener listener) - To register or unregister for incoming notifications via GCM or FCM via
Voice.register(Context context, String accessToken, Voice.RegistrationChannel registrationChannel, String registrationToken, RegistrationListener listener)andVoice.unregister(Context context, String accessToken, Voice.RegistrationChannel registrationChannel, String registrationToken, RegistrationListener listener). Once registered, incoming notifications are handled via aCallInvitewhere you can choose to accept or reject the invite. When accepting the call an access token is not required. Internally theCallInvitehas its own accessToken that ensures it can connect to our infrastructure.
As mentioned above, an access token will eventually expire. If an access token has expired, our infrastructure will return error EXCEPTION_INVALID_ACCESS_TOKEN_EXPIRY/20104 via a CallException or a RegistrationException.
There are number of techniques you can use to ensure that access token expiry is managed accordingly:
- Always fetch a new access token from your access token server before making an outbound call.
- Retain the access token until getting a
EXCEPTION_INVALID_ACCESS_TOKEN_EXPIRY/20104error before fetching a new access token. - Retain the access token along with the timestamp of when it was requested so you can verify ahead of time whether the token has already expired based on the
time-to-livebeing used by your server. - Prefetch the access token whenever the
Application,Service,Activity, orFragmentassociated with an outgoing call is created.
You can find more documentation on getting started as well as our latest Javadoc below:
To learn more about how to use TwiML and the Programmable Voice Calls API, check out our TwiML quickstarts:
- TwiML Quickstart for Python
- TwiML Quickstart for Ruby
- TwiML Quickstart for PHP
- TwiML Quickstart for Java
- TwiML Quickstart for C#
Please file any issues you find here on Github. For general inquiries related to the Voice SDK you can file a support ticket.
MIT