Research in Artificial Intelligence has not yet led us to AI "superintelligence," or the form of AI that we see in most movies and tv shows. Since these AI systems are not yet feasible, we must look for alternative solutions. Stanny works to mimic AI "superintelligence" by chaining together multiple algorithms with distinct features and capabilities.
The movie Iron Man is the reason I started programming. Back in 2013, I was determined to create my own version of Tony Stark's Jarvis. By the time I turned 18, I had taught myself enough computer science to start creating Stanny. The goal is to have Stanny capable of conversations similar to the "Mark II Test Flight" scene.
I developed one of the first iterations of Stanny as a WWDC Swift Student Scholarship submission. This form of Stanny contains a small, custom Natural Language Processing model and uses Apple's built-in speech-to-text framework to convert Stanny's responses into actual speech.
Three main processes make up Stanny: speech-to-text, natural language processing, and text-to-speech. The most important of these Stanny subsections is the NLP model. It uses a custom dataset to identify different speech "intents" and returns pre-written responses to the user. This model can function independently as an entirely text-based system.
An NLP model is not able to provide more than text-based responses, so Stanny uses an external action framework to interact with just about anything. Stanny can use the internet to access my projects and check the weather. If the device that runs Stanny is sophisticated enough, the program can utilize technologies like Bluetooth as well.
Humanlike text-to-speech algorithms are remarkably complex and require large amounts of data to function correctly. One of the best frameworks for developing these models is the Mozilla TTS library. It requires 20 or more hours of annotated speech and sophisticated hardware to train, but the results are well worth the effort.
Fill out my contact form with your question or comment and I'll get back to you as soon as I can!