Making a Monkey Thought Translator
Let’s invent something fun again! But this time, instead of something amazing but complicated like the Wearable Holodeck, we’re going to invent something that a hacker can make on a budget with off the shelf parts.
The Monkey Thought Translator! (Mark 2)
I freely admit it. This may go past ‘want’
I NEED my own Steve the Monkey
(perhaps not a monkey though … poo flinging and all that)
The Problem
The idea of a pet you can talk to is an old one. We see it frequently in cinema and literature but to date attempts to create any sort of voice/thought translator have been very limited and not very much fun.
Indeed, most pets only make a few sounds and simply translating those only gives us a few results . . . and we never really needed much help figuring out what those mean, did we?
We’ve even gone so far to try to create devices that use sensors to better gauge our pet’s mood, certainly a positive improvement and something that’s easily incorporated, but that’s simply a more advanced solution that approaches the problem from the same direction and only can be applied to a small number of very tolerant animals.
What many people really want is a way to properly anthromorphize their pet and have them respond to them entertainingly. We talk to our cats. We give our dogs human traits. We give our turtles cozies for their shells. We imagine them with the voices of Morgan Freeman or Mae West.
We want them to respond to us.
To solve this, we need to shift the context away from the pet and towards the owner.
Once we do, we open up a realm of amazing possibilities. Dug the Dog and Steve the Monkey aren’t unachievable with this approach, they’re early prototypes.
The Solution
By shifting the focus to the owner we get a new starting point . . . the electronic butler. Think Siri or Jarvis from Iron Man. We already have decent voice recognition that’s very trainable, and we can train it to respond to the owner and use the pet’s own voice (when available) to adjust that response.
You can talk to your bunny . . . and your bunny can talk back.
From there we can add a multitude of sensors (upgrades!) that obtain more information from the pet (basic things like heart rate, temperature, noises above/below human perception), from the environment (location, time and date, ambient environment.), and from the owner (the tone of your voice, what’s going on in your life, how late you are for work)
In addition, this approach is very learner-friendly. A simple phone app could be used to dynamically learn what sort of response you want from your pet at what times. Do you want your pet to be more apologetic in the morning? More motivational on Friday nights? A better listener when you’re having a bad day? An app with a few simple buttons can be used as a learner to slowly improve the device’s performance.
Early experiments using a Raspberry Pi and Jasper were very successful (well, for me) and the effect is powerful even when one is aware of the illusion (also tested with a simple microphone/funny voice/speaker setup). An Arduino may be sufficient processing power-wise but more may be useful as one adds sensors and attempts to interpret more inputs.
