Desh Raj

gss. A simple package for Guided source separation (GSS)

numerical-linear-algebra. Solutions to exercises in Numerical Linear Algebra by Trefethen and Bau.

dover-lap. Python package for combining diarization system outputs.

diarizer. Clustering-based methods for overlapping diarization

gmm-hmm-asr. Python implementation of simple GMM and HMM models for isolated digit recognition.

crnn-relation-classification. Tensorflow implementation of the Convolutional Recurrent Neural Network model with max pooling and attentive pooling, for relation classification on biomedical text.

pytorch-tdnn. Pypi installable TDNN and TDNN-F layers for PyTorch based acoustic model training

spyder. Simple Python package for fast DER computation

dnn-hmm-asr. Hybrid DNN-HMM model for isolated digit recognition

desh2608.github.io. Personal homepage for Desh Raj

attention_model_clinical_text. This code implements attention network on top of the CNN used in the paper titled "Relation extraction from clinical texts using domain invariant convolutional neural network" which was accepted in BioNLP at ACL-2016, Berlin, Germany

kaldi-noise-vectors. Implementation of different noise embeddings for noise aware training of Kaldi acoustic models.

css. PyTorch implementation of Continuous Speech Separation

kaldi. This is now the official location of the Kaldi project.

tweet-irony-detection. Model for the Sem Eval 2018 Task 3

cacgmm. GPU-based CACGMM trainer in CuPy

ovad. Overlapping speech VAD

beamformer. Souden MVDR beamformer on GPU with CuPy

alignment_restricted_transducer. An implementation of AR-RNNT loss as proposed in this paper: https://arxiv.org/abs/2011.03072

relation_extraction_in_clinical_text. CNN for relation extraction among problem, treatment and test in clinical text.

wpe. Nara-WPE in CuPy for GPU inference

gerrymandering. Course project for EN 553.732 Bayesian Statistics. We look at politicial gerrymandering of the 2016 Presidential election results in the state of Ohio from a Bayesian perspective.

pytorch-kaldi-neural-speaker-embeddings. A light weight neural speaker embeddings extraction based on Kaldi and PyTorch.

speech-datasets. Various speech datasets made available to the public

kaldiio. A pure python module for reading and writing kaldi ark files

spatial-transformer-networks-experiments. Source codes and result data for the Spatial Transformer Networks project for "CS 574 Computer Vision using Machine Learning" course offered in the Fall 2016 semester at IIT Guwahati.

pyannote-audio. Neural building blocks for speaker diarization: speech activity detection, speaker change detection, overlapped speech detection, speaker embedding

ai-deadlines. :alarm_clock: AI conference deadline countdowns

kaldialign. Python wrappers for Kaldi Levenshtein's distance and alignment code.

THULAC-Python. An Efficient Lexical Analyzer for Chinese

awesome-diarization. A curated list of awesome Speaker Diarization papers, libraries, datasets, and other resources.

old-academic-site. My personal academic webpage. Template borrowed from @OleVik

queue-utils. Utility scripts to submit multi-GPU jobs on CLSP, COE, and MARCC

iText. Simple, elegant text editor for Windows with Multiple Clipboard, Themes, Line Numbers.

transducer-loss-benchmarking. Python

constrained-optimization. Python based implementations of some popular constrained optimization methods.