Have you ever wanted to hear Clinton, Trump or Obama say something? Anything? We have. And now you can. We used cutting edge natural language and speech recognition technology to build a general purpose voice speech tool just in time for the 2016 Elections.
What it does
First and foremost, clicking "I'm Feeling Lucky" will allow anyone to instantly watch a hypothetical debate between Clinton, Trump, and Obama powered by Markov Chains generated from their very own speeches.
Furthermore, our project receives user text input (or generates our own sentence) and the user's selection of the politician, searches our database of words that the candidate has said and produces an audio file of Clinton, Trump or Obama speak the text input.
How we built it
We built our project by using advanced algorithms (and very powerful servers) to process audio files of Clinton, Trump and Obama speaking and save snippets of the files to put them together for the output desired.
Challenges we ran into
Sometimes, the words and phrases in the audio selections aren't spoken clearly or slowly enough. Furthermore, vocabulary is limited to the words that they have said. We tried breaking words down into phonemes for a larger library of speech but that didn't work too well unfortunately.
Accomplishments that we're proud of
Now we can finally hear Trump say non-discriminatory things (or hear Clinton say very discriminatory things)!
What we learned
Markov Chains are incredibly effective at predicting speech and lots and lots of super powerful hardware along with some clever processing is required for any reasonable level of speech comprehension.
What's next for HilarityTrumpton
Making America Great Again