The way it works is.... First and foremost, you need data annotators. You need people who can actually physically and manually give context to sentences, and then train the AI model to detect tone. You would take sample text. I gave the example “Today I'm having a bad day.” If you had a sentence like that, a person would need to give context. Ideally, you would want a young person who has more context, because kids have different language patterns and lingo. From there, that's how you would be able to train a model to detect tone.
On October 3rd, 2022. See this statement in context.