Post written by Felice Ho:

Fun with Music and NLP

Thank you to conference organizers and WiMLDS for the chance to attend PyGotham 2016 at the United Nations!

My favorite talks were related to music. Curiosity on how music streaming companies give users personalized recommendations led me to attend ‘Vector Space Modeling on Music Data’ by Tim Schmeier. We learned about one company’s approach using a mapping of entities (i.e. artists, songs, albums, etc) into high dimensional space. Distances between two entities correspond to their similarities or differences. Schmeier briefly discussed acoustic modeling and the use of deep learning to teach computers how to distinguish between songs that sound similar and those that don’t. This talk quickly showed how recommendations work not only in digital music but also in other applications such as Netflix or Amazon.

I also enjoyed ‘The Sound of Data: Using Python to Transform Data Streams into Music’ by Gabriel Levine. Instead of visualizing data into charts and graphs he chose to transform streams of data into sounds and music. Levine gave a live demo using recent weather data from Manhattan. Although we all laughed about how eerie the current weather sounded, I had a lot of fun at this talk and loved ‘listening’ to data!

A talk I missed out on but hope to watch online is ‘Making Sense of 100 Years of NYC Opera with Python’ by Suby Raman. I heard this was an excellent talk about the scraping and analysis of data from the Metropolitan Opera of New York. Raman uncovered surprising findings about American art culture.

There were a number of talks on natural language processing so I ended up attending several including ‘Summarizing documents’ by Mike Williams, ‘Everything You Always Wanted to Know About NLP but Were Afraid to Ask’ by Steven Butler and Max Schwartz, and ‘Higher-level Natural Level Processing with textacy’ by Burton DeWilde.

Having used Python for web development so far and with no prior exposure to machine learning or artificial intelligence, PyGotham gave me the opportunity to explore the amazing things we can do with computers and learn some tools to do it with Python! I have newfound fascination for the possibility of teaching computers to think as we do and for computers to perform as we want them to.
Lastly, as someone who turned to programming from a non-engineering background, I loved that so many of the speakers come from different occupations. Talks listed above include linguists, musicians, and physicists! I love that code creates the opportunity to bring together people that do not normally work together but can find ways to connect.