How to Automate Conversational UI and Voice Interface Testing
Say Hello to Conversational UI
If you aren’t already talking to your watch, your car, or your smartphone, you will be soon. Users love interacting with services and content on their own devices, at their convenience and on their terms- doing so in a more “human” way frees them inconvenient and tedious text entry. Voice interfaces can make commutes more productive, let you quickly find a TV show, or find great recipes without having to juggle a keyboard and mouse with messy baking hands.
As a result of this trend, brands are adopting conversational interfaces as a key element of their digital strategy, expanding their presence beyond web browsers and native mobile applications. Ally Bank, for example, already offers a set of “Alexa skills,” including balance checking, money transfers, recent transactions, etc. The growth in developing Alexa skills has been remarkable. They are becoming an increasingly ubiquitous tool.
What to Test
When your app is utilizing Google Home, Alexa, etc., the first question to ask is “what do I test?” Well, at least for conversational interfaces, you should count on hardware manufacturers to offer a unified voice-to-text/text-to-voice interface. One would hope that each bank won’t need to test every Google and Alexa device variant in the market!
Instead, brands should focus on the functional and responsiveness testing of the service itself. For example, are bank transactions executed correctly? Can users order the correct product or listen to the right music?
While testers can leverage simulators and APIs offered by Google and Amazon for these purposes, there is a small twist making unified testing a little trickier: Alexa and Google Assistant on mobile phones. Of course! Why not allow users to access their Amazon retail experience with Alexa when they aren’t home or away from the Alexa device?
How to Test Conversational UI
Amazon recently introduced Alexa applications for Android and iOS. Google also offers access to Google Assistant on these platforms. As a result, a tester can (and in this case must) test these services on mobile devices.
Using Perfecto’s audio testing solution, it’s not only possible to test conversational chatbots, but also Alexa, Google Assistant and Siri. Our solution offers both injection of input from a string (converted to audio and injected to the device), or from a pre-recorded audio file, enabling diverse sets of test conditions, such as background noise, multiple speakers, etc.
On the validation path, one can validate the function and responsiveness via:
- Native or visual object validation, in which case the interface responds both visually and vocally
- Recording the audio and converting to text
- Validating the audio itself
Turning the Lights On in Nick's Living Room
In this example, Nick turns on Alexa/Google Home, clicks on the input button and orders it to turn on/off the lights. On Google Assistant, it’s even possible to see the response.
Get Ready for Conversational Interfaces
Conversational interfaces have already proliferated throughout many aspects of our lives: from watches, smartphones, and TV remotes to digital personal assistants - with other devices coming quickly to market - cars, kitchen tools, vacuums. As experiences and services expand, it will be critical to offer repeatable and consistent test automation for these interfaces. Luckily, there’s a way to do that today!
Learn more about testing emerging technologies. Check out our article, IoT Testing for BLE Protocol.
- How AI Will Change Test Automation in 2019 & Beyond [eBook]
- The Importance of 5-Star App Ratings [blog]
- How to Scale Up Your Test Automation Practices [eBook]
Many thanks to Nick Sanjines for the awesome example and video. Check out Nick on Linkedin:
Sr. Solutions Architect at Perfecto Mobile