Training Noise-Robust Spoken Phrase Detectors with Scarce and Private Data: An Application to Classroom Observation Videos

We explore how to automatically detect specific phrases in audio from noisy, multi-speaker videos using deep neural networks. Specifically, we focus on classroom observation videos that contain a few adult teachers and several small children (< 5 years old). At any point in these videos, multiple...

Full description

Bibliographic Details
Main Author: Zylich, Brian Matthew
Other Authors: Gillian Smith, Reader
Format: Others
Published: Digital WPI 2019
Subjects:
Online Access:https://digitalcommons.wpi.edu/etd-theses/1289
https://digitalcommons.wpi.edu/cgi/viewcontent.cgi?article=2288&amp;context=etd-theses