Daydream VR from Google was released late 2016. It was a low-priced approach to a virtual reality entry by Google. It was not a coincidence that the first Google Daydream VR capable phone happens to be the Google Pixel and Pixel XL. Now the time has come to release VR games to have the content needed to fuel VR sales.
Voice Command on VR
One of the latest games is called Untethered. It was made by Numinous Games, and it adds its unique approach to VR: voice commands. Google is going big for voice commands, as well as AI, and it seems fitting that one of the earliest games on Daydream VR is voice enabled. You navigate through the game using voice commands. That is so cool!
Traditional games make use of escape codes and menus. This method was adapted by VR games, in a natural manner. You click the menu with your hands, and you see all of this through your VR glasses as a floating menu. Others, of course, will see you clicking on invisible buttons in the air in front of you. After a while, it stops being awkward and feeling unnatural.
The use of a voice command, however, is more natural. Anyway, you are talking to yourself when you play so that the voice command would be a logical progression.
Untethered, the Episodic Game
The main idea of Untethered is that it is episodic. Things unfold as you continue with the game. You get to begin new episodes when they are released. In the meantime, you go through the game again and see different results, or you do a do-over. Either way, the game becomes easier, and it feels more natural as you play it again and again.
For each episode, the same story is told but from a different perspective. In the first episode, you are a radio DJ. You talk to your listeners as they give you a call while you are on the air. Through these interactions, the story unfolds, and you get a better idea of what you are supposed to be doing, and what the mystery and game are all about. The next episode occurs at the same time as the first episode, but this time you are a different character. So far, the second episode has not yet been released, and there is still no idea what draw you would play.
The episodic nature of this game is inherent in most games, what is different is that you play different characters in each episode, and the game always starts at the same time setting. Think of it as Rashomon, with multiple points of view of the same story. The repetitive nature of the storytelling would give the player an idea of what is happening in the story. He can also evolve in his line of questioning, building upon what he has already learned in the first episode. I would not be surprised if the radio show is playing in the succeeding episodes while you play the game. It would be great to hear your recorded voice from the first episode as you go through the rest of the game.
The player would be interacting with actors built into the game. The action would be event-driven according to prompts from these in-game characters. This forces more interaction between player and the in-game characters. He should try to talk to them as much as he can.
Voice Commands and Its Issues
Voice commands have been under development since the first HAL 9000. Or even earlier. Star Trek had voice commands on the Enterprise computer. However, the problem with voice commands is that the recipient (the computer) is expecting a fixed way of answering, or close enough. It has parameters to follow the voice. However, the reality is that most people talk differently from whatever parameters are in the voice parser. That means that people who speak slowly, have an accent, a lisp, or any other voice problem, will have a hard time communicating with the computer.
Even the best voice dictation software hover between 80 to 90 percent correct. They usually start as low as 60 percent correct. With constant use, the AI behind the voice recognition and parser would get to understand the individual speaking. There’s also the additional problem of having multiple players, or speakers. When you start calibrating for one person, you would have to explicitly tell the voice recognition software that a different person is using the account. In which case, it would have to learn to decipher the new person’s voice pattern all over again.
AI and Voice Recognition
This would explain why AI, at present, has tightly meshed with voice recognition. The responses of the AI are dependent on the context of the question. The AI could not understand context unless the speaker keeps on talking. The more he talks, the more the AI learns.
Untethered, the game, would be hard put to handle any player with a voice or speaking problem. It cannot handle stuttering, nor can it handle soft voices. This is not just a problem of the game but is the current state of AI.
The premise of Untethered is laudable, and it should get a wide audience, if only because of the script. A game with a voice command, as well as expected voice interaction between the player and the AI, should be a fun game. You would expect a witty repartee, however hard that may be to do. All in all, this is something to look forward to.
Google+