Sorry, you need to enable JavaScript to visit this website.

UNSUPERVISED TRAINING OF A DEEP CLUSTERING MODEL FOR MULTICHANNEL BLIND SOURCE SEPARATION

Citation Author(s):
Lukas Drude, Daniel Hasenklever, Reinhold Haeb-Umbach
Submitted by:
Lukas Drude
Last updated:
10 May 2019 - 12:34pm
Document Type:
Poster
Document Year:
2019
Event:
Presenters:
Lukas Drude
Paper Code:
1819
 

We propose a training scheme to train neural network-based source separation algorithms from scratch when parallel clean data is unavailable. In particular, we demonstrate that an unsupervised spatial clustering algorithm is sufficient to guide the training of a deep clustering system. We argue that previous work on deep clustering requires strong supervision and elaborate on why this is a limitation. We demonstrate that (a) the single-channel deep clustering system trained according to the proposed scheme alone is able to achieve a similar performance as the multi-channel teacher in terms of word error rates and (b) initializing the spatial clustering approach with the deep clustering result yields a relative word error rate reduction of 26 % over the unsupervised teacher.

up
0 users have voted: