How We Built a Crime-Solving Simulation Game Powered by Voice
We used state-of-the-art technology to create a new gaming experience
Building for voice first is hard. Perhaps that is what makes it so compelling. Building engaging voice first games is exceptionally difficult, mainly because you cannot rely on flashy visuals.
About 2 years ago we had a crazy idea to build a voice-first crime solving simulation game for Alexa. Fast forward 2 years, and we scrapped the initial design and restarted from scratch. Why — because our initial approach was already outdated.
For the new approach, we decided to harness the power of Alexa’s dialog capabilities. Our goal was to have an interactive, realistic dialog flow. This requirement dictated our use of a manual dialog strategy. When using a manual dialog strategy, it is up to the developer to control the flow of the conversation by explicitly eliciting responses from the user of the Alexa Skill.
To make this work we relied on the dialog model, entity resolution, handler logic, and most importantly, creativity. Although we were manually eliciting slots from the user, we knew it would not be feasible to also manually evaluate the spoken utterances from the user. In fact, I would strongly recommend against this in your skills. Instead, take advantage of the interaction model, the dialog model, and entity resolution.
As you can see above, our entire approach relies on the the slots and intents in the interaction model. We used custom slots with entity resolution as the core game mechanic. This enabled us to control the flow of the game by simply evaluating entity resolution matches or failures.
Hopefully, this gets your creative juices flowing. Voice first design is truly the next paradigm shift, and we are in the very early stages. This skill is just one of many innovative approaches to voice first experiences. If you’re interested, you can enable Signs of Crime for free, and check it out on any Alexa enabled device.