Overview / Usage

Raw audio is captured from the microphone, then converted it into to a visual representation of spectrum of different frequencies present in the sound, which is a spectrogram. Thin slices of the spectrogram are then fed into to a Recurrent Neural Network model as continuous time steps. This model is trained to predict the likelihood of a character being spoken in that time frame. Putting all together such likelihoods in a sequence gives us the textual form of the words spoken in the audio input, thus converting speech to text.

This PC voice-assistant skill supports five types of actions as voice-commands:

1. Game tactics –
   Using the command “high pressure” to change tactics of how the team plays the match

2. Skill and moves –
   Enabling the “rainbow flick” skill in the game by giving voice command “flick right”

3. Goal celebrations –
   Using the command “work out” to perform a unique goal-scoring celebration.

4. Substitute Players –
   Using the command (for e.g.) “Substitute Frabegas for Kante” to perform a quick substitution.

5. Changing Formation –
   Using Commands such as “Change formation to 3-5-2” to change formation structure to 3-5-2 of the team.

Implementing this PC voice assistant Skill will not only simplify controls in the PC games, but also act as a perfect complementary input method for PC gaming enthusiasts as it can be integrated with any PC games of the user’s choice.

Methodology / Approach

This project uses Amazon Web Services and Alexa Skills Kit. It is also uses .NET framework via Microsoft Visual Studio on the Intel-based development machine.

Technologies Used

Intel i7 8th gen
Intel HD graphics 620
Alexa Skills Kit
Amazon Web Services
MS Visual Studio 2017

Other links

• Enabling the “rainbow flick” skill in the game by giving voice command “flick right”