Jacob Kahn

Jacob is a Research Engineer at Facebook AI Research (FAIR), focusing on speech and language. He works on speech recognition methods, parallel and distributed architectures and optimization for deep learning. Jacob holds M.S. and B.S. degrees in computer science from the University of Pennsylvania’s Management & Technology program, and a B.S. in economics from UPenn's Wharton School, where he focused on statistics and operations research.

Jacob's Publications

January 13, 2020

RESEARCH

NLP

Scaling up online speech recognition using ConvNets

We design an online end-to-end speech recognition system based on Time-Depth Separable (TDS) convolutions and Connectionist Temporal Classification (CTC). The system has almost three times the throughput of a well tuned hybrid ASR baseline…

Vineel Pratap, Qiantong Xu, Jacob Kahn, Gilad Avidov, Tatiana Likhomanenko, Awni Hannun, Vitaliy Liptchinsky, Gabriel Synnaeve, Ronan Collobert,

January 13, 2020

May 04, 2020

SPEECH & AUDIO

NLP

Libri-light: A benchmark for ASR with limited or no supervision

We introduce a new collection of spoken English audio suitable for training speech recognition systems under limited or no supervision. It is derived from open-source audio books from the LibriVox project. It contains over 60K hours of audio, which is, to our knowledge, the largest freely-available corpus of speech.…

Jacob Kahn, Morgan Rivière, Weiyi Zheng, Evgeny Kharitonov, Qiantong Xu, Pierre-Emmanuel Mazaré, Julien Karadayi, Vitaliy Liptchinsky,Ronan Collobert, Christian Fuegen, Tatiana Likhomanenko, Gabriel Synnaeve, Armand Joulin, Abdelrahman Mohamed, Emmanuel Dupoux

May 04, 2020

April 24, 2020

RESEARCH

NLP

Self-Training for End-to-End Speech Recognition

We revisit self-training in the context of end-to-end speech recognition. We demonstrate that training with pseudo-labels can substantially improve the accuracy of a baseline model.

Jacob Kahn, Ann Lee, Awni Hannun

April 24, 2020