I’m struck by how different this project might have turned out if I’d made some divergent choices along the way. I briefly played with other forced aligners besides Gentle that return different kinds of information. Darla, for example, gave me results with the formants for each vowel. Some forced aligners output the data in TextGrid files, which are used primarily with Praat, a software tool for analyzing phonetics. I’m sure those files can be accessed by Python just like CSV or JSON files, but the different experience of working with them may have thrown up other possibilities and obstacles. Before I began this independent study I had a vague inkling of where I wanted to go, and that it should be possible, but I didn’t know enough about this area to make informed decisions about the best path. Johanna Devaney’s guidance was crucial throughout the semester.

There are other Max/MSP libraries besides MuBu that deal with organizing a large corpus of sound, such as FluCoMa. I plan to look into those a bit more and may incorporate multiple approaches into my Max patch.

In a few weeks I’ll meet up with my collaborator Aaron Snyder to put all of this work to compositional use. That will really be the test.