PitchShift

Obstacle Course (Loss)
Opening Screen

Inspiration

The inspiration for this game comes from wanting to use either facial or voice recognition as the foundation of the project. The games that mostly influenced this idea were Guitar Hero, games with voice chat, and more! It also came to mind that this game could possibly help some people who have speaking or pronunciation problems when it comes to words. All in all though, this game is meant to be challenging but very fun!

What it does

PitchShift has players use their voices to play multiple game modes. The game modes include:

Replay: Player(s) say the words shown on the screen and earn points for each word they get correct. The difficulty increases with each round.

Obstacle Course: Players use their voice in order to avoid obstacles to complete levels. The difficulty increases with each round.

How we built it

The game was mostly built using Unity software. SQLite Database was used to store in words meant for a mode in the game. A lot of YouTube tutorials, googling, and troubleshooting was involved in making this project. The team was separated so that we could cover each main category of what makes a game special. Collectively we managed to find and create sprites, create cool UI, find good gaming music, put in voice recognition, and the list goes on!

Challenges we ran into

After incorporating voice recognition into the game, we realized that the computer will always take around 2 seconds before picking up on a user's input, even if it's just one word. This makes the game much slower and far too easy for a specific mode. Our solution to this problem is to add more words for the user to say to make up for the lack of speed.

Unity's DictationRecognizer (the code that allows the computer to tell what the player is saying) throws an error and stops working after 5 seconds of the user not speaking into their microphone. Finding a solution to this was incredibly difficult because seemingly no one has solved it and this error would not be able to work in a built game unlike in the editor.

Came across an issue where reading a txt.file required reading it at a specific location which wouldn't be on all computers. To fix this, the SQLite database was used to retrieve words ranging from "Easy - Hard". This method should work for all PCs.

Unity's editor build can be deceiving because the actual built version throws many errors. For instance, I tried accessing a database created that stored many words but I was given many many MANY errors.

Time managing was incredibly difficult. Ideas had to be dropped because they were very complicated to do within 24 hours, such as "Spelling Bee", a mode where it was intended that a chatbot would say the word to the user and then the user would have to spell it out to earn points.

Accomplishments that we're proud of

1 Incorporating voice recognition into the game\n 2 Creating multiple game modes 3 Having a good color scheme

What we learned

We learned how long and what it takes for AI to recognize words. In the case of how Unity does it, Dictation Recognition goes through several (often inaccurate) hypothesizes before making a final guess. And even then, that final guess can be rejected if it doesn't match a certain standard. Some of us learned how to use Unity UI and draw sprites with asperite.