This project seeks to sonify bias in word embeddings present in word2vec. Using word2vec2osc, Wekinator, ChucK and Processing, a machine learning model translates words that are more female or more male into an audiovisual experience. All analogies are real constructs of word2vec trained on Google News. The poles of male and female are realized as tonal centers, while the hectic sine lead voice represents the degree to which words that should be gender-neutral are not.