Abstract
Speeches represent an important communication format for a wide range of government, business, and social activities. However, evaluating speeches is entirely opinion based today and has a lot to do with our feelings about a speaker as well as the views of those commenting on the speech in positions of power.
So what if there were a way to improve the effectiveness of a speech by looking solely at the words, phrases, and sentences that made it up?
We set out at Treegoat to determine if a deep learning artificial intelligence (AI) model could provide a statistical method of scoring how “interesting” a speech was based solely on its text. Could we train an AI model to learn what makes a good speech beyond the trappings of its speaker, topic, and occasion?
You may ask, what would be the point? Aren’t speeches made to be heard and not read? Doesn’t an analysis that considers just the text fail to consider its medium and our expectations?
That is exactly the point. Our AI model isolates the content alone, leaving the subjective dimensions of style and context out of the analysis. Might having an added point of analysis for a speech support its likelihood of success?
The results were unexpected.
Receive your complimentary report of 25 pages, including over 30 charts and graphs evaluating the output.
Get the White Paper!
By requesting the white paper, you agree to receive marketing emails from Tree Goat Media, Inc. We respect your privacy and you may unsubscribe at any time.