Joseph Turian's Projects
Deepo-based audio docker
Code for the paper Hybrid Spectrogram and Waveform Source Separation
Demucs Lightning: A PyTorch lightning version of Demucs with Hydra and Tensorboard features
Implementation of DiffWave and SaShiMi audio generation models
IM-like application for Pinax social networks (Django), that allow you to see which friends are online and chat them
django-lazysignup is a package designed to allow users to interact with a site as if they were authenticated users, but without signing up. At any time, they can convert their temporary user account to a real user account.
Dmoz RDF parser
Extract faces from video clips; generate training data for pose-invariant face features
Django project to do blind testing and figure out which of your friends post things you actually like
Render dx7 patches, dockerized
Retrieve text embeddings, but cache them locally if we have already computed them.
State-of-the-art deep learning based audio codec supporting both mono 24 kHz audio and stereo 48 kHz audio.
fabric recipes, primarily for deploying Ubuntu and EC2 instances.
fast-stable-diffusion, +25-50% speed increase + memory efficient + DreamBooth
Deploy FatFree CRM on EC2
flair/pos-english tagger, as a replicate.com model
Fork of FLANN 1.2, Fast Library for Approximate Nearest Neighbors
flickorpus collects an image and tag corpus from flickr.
Grab all Wikipedia abstracts, in all languages
Grok your music collection, and save it into a persistent format.
HTML5UP Directive, as a Jekyll theme
Inverse audio synthesis
Jekyll theme with beautiful LaTeX typography
KEA 5.0 (keyphrase extraction software), modified to be an XML-RPC service
Minimal keyword extraction with BERT
Technical blog
A neural language model, intended to produce embeddings for a linear classifier