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

HIGH-QUALITY SPEECH CODING WITH SAMPLE RNN

Primary tabs

Citation Author(s):
Janusz Klejsa, Per Hedelin, Cong Zhou, Roy Fejgin, Lars Villemoes
Submitted by:
Lars Villemoes
Last updated:
23 May 2019 - 7:33am
Document Type:
Poster
Document Year:
2019
Event:
Presenters Name:
Janusz Klejsa
Paper Code:
SLP-P24.7

Abstract 

Abstract: 

We provide a speech coding scheme employing a generative model based on SampleRNN that, while operating at significantly lower bitrates, matches or surpasses the perceptual quality of state-of-the-art classic wide-band codecs. Moreover, it is demonstrated that the proposed scheme can provide a meaningful rate-distortion trade-off without retraining. We evaluate the proposed scheme in a series of listening tests and discuss limitations of the approach.

up
0 users have voted:

Dataset Files

Audio demo

(192)

Poster

(194)