Extracting pain points from respondents feedback

I need some help for my personal research where I am trying to analyze free text customer responses that I have scraped from web to get an idea around what are the Themes and Customer pain points derived from in my datasets. I am just looking to get two columns from the feedback text as I explained above. I want someone to help me with this using state of the art NLP techniques.
The critical element is that the outcomes in both the columns i.e. "Feedback Themes" and "Customer pain point" should be humanly readable and not merely machine generated text. The coding should be done in Python(using Jupyter Notebook) so that I can adopt the same and build on top of it.