Speech separation with deep neural networks

We are sorry, this position has been filled.

The project aims to use the latest advances in deep learning to tackle the problem of speech separation from one or more microphones. The goal of the project will be to produce separated audio from a mixed input using high dimensional embedding and time-frequency masking methods. 

Lab: Neural acoustic processing Lab

Direct Supervisor: Nima Mesgarani

Position Dates: 5/27/2019 - 9/1/2019

Hours per Week: 20

Paid Position: Yes

Credit: No

Number of positions: 2

Qualifications: Deep learning, speech and audio signal processing

Eligibility: Master's (SEAS only)

Nima Mesgarani, nm2764@columbia.edu