Inspiration

The Skyrim Alexa skill (available only in English) was a revelation for me about the potential of the vocal games, allowing to play to a video game... without the video, and so strongly relying on the player imagination. But this skill will probably never reach the French language... so I though I had to try to make it myself. The idea was to manage providing an experience which would never be the same. Like mix of a vocal Minecraft and a vocal Skyrim. The goal was high, and the current status is more a strategy automatically generated waze, but I hope adding feature to reach the initial goal.

What it does

The current status of the game generate randomly a waze forest, each place of this waze being a different combination of several criteria. This way the user feel like walking in a forest where each place is unique, with its specificities. Some places are dangerous, some others seems like but are not. The user can look around before moving, to avoid loosing, but is limited for this action. When loosing, he can restart in a full new forest, or restart in the same forest to try improving his score. To make it more real and fun, I choosed to put Alexa in the forest, the player is only giving instructions to Alexa, who can refuse or comment the user actions which are sending her to a painful death.

How I built it

The skill is based on AWS lambda usage, without specific additional API. That was my very first Alexa skill. So I had to discover everything, and to put in place a structure a split in a clean way the different aspect of a skill: intents, handlers, states, utils, language, variables...

Challenges I ran into

The full skill was a challenge by itself, but several points were harder than others: the generation of the waze insuring to always have the possibility to exit for the user, the managements of all built in intents (overall reprompt), and making Alexa speech natural and not repetitive.

Accomplishments that I'm proud of

Alexa will never say 2 times the same thing the same way. All of the hundred of sentences have 3 or more variants, making the discussion really more natural. This needed hours of writing (and testing, to ensure the prononciation of each sentence, or add speech effects).

The management of the waze generation is also a thing I'm proud of, with the possibility to vary its complexity or its size.

What I Learned

I discovered the Alexa node API, and some aspects of JavaScript (that I didn't know so much).

What's next for Perdu en foret

The potential of this game is big, and I'm thinking about lot of improvements, like adding monsters, villages, adding something outside of the forest, perhaps adding in skill purchases to access some of these new features...

Try it (French only)

https://www.amazon.fr/Guillaume-Pelouas-Perdu-en-ForĂȘt/dp/B07G8JV9SP/ref=nodl_

Share this project:

Updates