Surface tap localization - Python scripts for determining the location of a tap on a surface via an array of contact microphones.

PPM in Swift - Prediction by partial matching (PPM) language model implemented in Swift.

HoloLens surface keyboard - HoloLens v2 keyboard for investigating typing on a surface co-located keyboard.

HoloLens midair keyboard - HoloLens v2 midair keyboard for comparing index and ten-finger typing with and without eye-tracking based filtering.

Audio collector - A Java application to record spoken dialogue turns in a two-person conversation. The application can recorded up to four microphones simultaneously.

Text predictor for Python - Makes word and character predictions based on an N-gram language model.

Baton - App that allows users of augmentative and alternative communication (AAC) to donate selected things they've written with their AAC system.

Language models for AAC - Various language models optimized for conversational style communications. This includes both n-gram and large language models (LLMs).

HoloLens typing - Additional data and detailed participant log files for two typing studies we carried out on a HoloLens v2 headset.

Surface tap localization - Audio of people tapping on a table instrumented with eight contact microphones.

Surface tap audio - Audio of people tapping on a QWERTY keyboard printed on a table. Audio recorded by eight contact microphones.

Surface tap videos - Videos of people typing on a QWERTY keyboard located on a wall. Videos are from an egocentric (first person) perspective recorded by a phone near a participant's forehead.

SpokenJava - A dataset for investigating automatic speech recognition in the domain of spoken programming languages.

Nomon single-click data - Click data from motor-impaired users collected in our ASSETS 2023 study on the single-switch interface Nomon.

Language model personalization dataset - Contains resources we used in this paper to conduct language model adaptation experiments based on the Enron Personlization Validation Set.

Noisy typing on QWERTY keyboards - Contains typing data from participants in nine previously published text entry studies. Data was collected using a variety of devices (touchscreen phone, smartwatch, mid-air VR/AR keyboard, desktop keyboard), and using different keyboard features and user input strategies.

Mobile text dataset and language models - Text written on mobile devices mined from web forums. Also includes a variety of trained language models.

Challenging Twitter phrase set - Sentences from twitter designed to be challenging to recognize. 213 out-of-vocabulary phrases, 194 in-vocabulary phrases.

Enron mobile email dataset - Genuine mobile email sentences written by Enron employees on their BlackBerry mobile devices.

COMM2 test set - A test set of AAC-like communications

Conversational AAC resources - Text corpus and language models for augmented and alternative communication (AAC)

Big word lists - English word lists with 22K to 1699K unique words.

Cross-platform recording - Recording from the microphone using cross-platform code.

Cross-platform recognition - Speech recognition on live audio using Sphinx-3 and cross-platform code.

SRILM on Windows - How to build SRILM on Windows using Visual Studio.

Composition task resources - Various files used in our paper on composition in text entry evaluations.

Turk dialogues - Dialogues invented by Amazon Mechanical Turk workers.

Vocab 100K - English word vocabulary of 100K words.

Nomon - Interface for communicating via a single switch.

Technology Assisted Conversations - Resources for improving Augmented and Alternative Communication (AAC), including trained langauge models optimized for AAC-like text.

Parakeet - Mobile text entry using speech and a touch-screen interface.

Speech Dasher - Text entry using speech and navigation by pointing.

HTK WSJ - HTK acoustic models using Wall Street Journal corpora.

CMU Sphinx WSJ - Sphinx acoustic models using Wall Street Journal corpora.

CSR - Language models using 222M words of text from the CSR LM-1 corpus.

Gigaword - Language models using 1200M words of text from the English Gigaword corpus.

SPH fluid flow - Modeling fluid flow using smoothed particle hydrodynamics.

LaTeX bar charts - Create LaTeX tables containing both data and bar graphs.

Java / C++ sockets - Classes to allow socket communication between Java and C++.

Stopwatch - A simple free stopwatch program.

3DPath - Utility to help create 3D spline camera animation paths through a ray-traced scene.

NASA TLX - Implementation of NASA task load index in HTML + JavaScript