Arlie Coles

Pursuing machine learning solutions to natural language problems.

I'm Arlie Coles, a Professional Master's student studying machine learning with MILA at the Université de Montréal. I am there to deepen my knowledge and learn about the cutting edge of the field so I can bring creative and practical contributions to the growing machine learning industry.

My strong background in NLP and interest in ASR were cultivated at McGill University, where I studied computer science and linguistics and recieved my B.A.&Sc. with First Class Honours. I developed corpus-analysis software for the Montreal Computational & Quantitative Linguistics Lab and developed neural networks for the Montreal Forced Aligner for my honours thesis under Morgan Sonderegger.

Skills & abilities

  • Programming: Python, Java, C/C++, Lua
  • Scripting: Bash, R, Perl, PHP, Javascript
  • Design: HTML, CSS, Photoshop, LaTeX
  • Tools: Unix, Git, Docker, TravisCI, Excel
  • Frameworks: JQuery, AngularJS, Django
  • ML/NLP/ASR: Pytorch, Keras, Kaldi, Praat

Recent projects

Forced Phonetic Alignment by Neural Network

Integration of Kaldi's nnet2 deep neural networks into the Montreal Forced Aligner, permitting the automatic forced alignment of speech audio and a corresponding transcript in time using a Deep Neural Network-Hidden Markov Model architecture.

GitHub | The Montreal Forced Aligner

Deep Rhyme Detection

Command line tool for the semi-supervised detection of rhyme and rhyme schemes (in the context of poetry or lyrics), including internal and slant rhyme, using bidirectional LSTMS.


Speaker Accent Classification

Three models for the automatic classification of speaker accent (UK vs. US English) given audio recordings, using several varieties of RNNs and CNNs.