[hacky_engineer]

I was working on a NLP(ish) idea for a while and got a descent prototype put together, but haven't had time to take the project past that. The idea was to train a model to classify tone of voice (a kind voice, angry voice, whiny voice, etc). Writing this out again now. I wonder if we could train a model to predict whether a statement is sarcastic or not..

[JPLeRouzic]

https://en.wikipedia.org/wiki/Sentiment_analysis

https://www.kdnuggets.com/2016/05/voice-tone-analysis-emotio...

[sharemywin]

The dataset would be the hard part.

I played around with Sentence Embeddings and Sentence Similarity.

I thought about being able to compare a sentence with "Happy" and get a score between them. This could be a generalized solution for a nlp problems.

I noticed it tended to match things a little different than I would expect. I would expect the subject of the sentence to match first.