Ken Schutte

Bus travel from Lima to Rio de Janeiro

In 2025, I traveled from Lima, Peru to Rio de Janeiro, Brazil by bus (as well as places along the way). Here are some details for anyone interested in doing the same.

Update to Apple TV App for Democracy Now!

I updated my Democracy Now! app for Apple TV. In addition to the feed of latest shows, you can now browse the whole archive.

BlueSky library-bot

Using the BlueSky API to create a bot that posts images of random libraries around the world.

ARC Viewer

I set of pages to quickly view many tasks from the ARC-AGI Challenge dataset.

Procession maps for Guatemala City, 20024

I created some custom maps showing the routes of Lent/Holy Week processions in Guatemala City

swap_8_and_9: How a simple import can modify the Python interpreter

Summary : I present a simple Python module that after imported, changes the values of two integers. The purpose is (a) show that modules can do harmful and unexpected things, (b) show an example of how to write a simple module in C, and (c) introduce a few things in CPython source code.

"Gzip beats BERT?" Part 2: dataset issues, improved speed, and results.

Summary : Update to my last post analyzing the gzip+knn paper. Some of the datasets have heavily contaminated train/test tests. I show some ways to significantly speed-up the gzip method. I present more results of my "fair" reimplementation.

Did Musk steal his new X logo?: glyph analysis in Python

Summary : I show how to use Python to analyze glyphs from SVG and fonts. I compare the new X.com logo to the corresponding glyph in a few different fonts. There are no exact matches, but some features closely match those in the Noto Font.

Learning PostScript by comparing it to Python

Summary : I provide a brief introduction to the PostScript programming language. I show a PostScript program to generate the Chicago flag and show a line-by-line translation into Python

Bad numbers in the "gzip beats BERT" paper?

Summary : This is an analysis of the paper Low-Resource” Text Classification: A Parameter-Free Classification Method with Compressors by Jiang et al. I analyzed their source code and found that their gzip + knn method is scored in a way that artificially inflates the accuracy in comparison to the baselines. It marks an example correct if either for the top two neareset neighbors is correct. I…

IPostScript: jupyter kernel for the PostScript language.

2023 Guatemalan election: in-progress data analysis.

Summary I've been doing some data analysis on the 2023 Guatemalan elections. In progess...

Phingerprint: An experiment in audio-fingerprinting on 4,240 hours of live Phish recordings.

Summary This page describes the process and results of applying audio-fingerprinting to finding snippets of audio in a large database of live concert recordings from a single artist. My goal was to search a database of 4,240 hours of concert recordings (roughly every circulating show by the band Phish) for a given "query" audio snippet. The queries are different encodings and often from different…

2023 Eurail trip: review coming soon.

Summary : This is a map of a trip I took in 2023 using the Eurail Pass. I am planning on writing up a review of the pluses and minuses of this rail pass.

Apple TV App for Democracy Now!.

I've written an app for Apple TV (i.e. tvOS) to watch the independent news program, Democracy Now! . The app is free. Me and this app are unaffiliated with the show. If you like it, make a donation to the show . Search for it on your Apple TV. Here's a link to its App Store page .

MIDI processing in MATLAB.

Summary : I've written some MATLAB / GNU Octave scrips to read and write MIDI files. Included are some very basic synthesis functions, which can easily be replaced by your own code. MIDI can be also be generated from note information you create within matlab: perhaps useful for audio transcription or algorithmic composition.

PhD Thesis: Parts-based Models and Local Features for Automatic Speech Recognition.

Summary : This page gives some information on my PhD thesis, Parts-based Models and Local Features for Automatic Speech Recognition, which was completed in May 2009. It was done at MIT in the SLS (Spoken Language Systems) group within CSAIL (Computer Science and Artificial Intelligence Laboratory). My thesis advisor was Jim Glass.