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

facebooktwittermailshare

On the analysis of training data for wavenet-based speech synthesis

Abstract: 

In this paper, we analyze how much, how consistent and how accurate data WaveNet-based speech synthesis method needs to be abletogeneratespeechofgoodquality. Wedothisbyaddingartificial noise to the description of our training data and observing how well WaveNet trains and produces speech. More specifically, we add noise to both phonetic segmentation and annotation accuracy, and we also reduce the size of training data by using a fewer number of sentences during training of a WaveNet model. We conducted MUSHRAlisteningtestsandusedobjectivemeasurestotrackspeech quality within the conducted experiments. We show that WaveNet retains high quality even after adding a small amount of noise (up to 10%) to phonetic segmentation and annotation. A small degradation of speech quality was observed for our WaveNet configuration when only 3 hours of training data were used.

up
0 users have voted:

Paper Details

Authors:
Zdeněk Hanzlíček, Jindřich Matoušek
Submitted On:
13 April 2018 - 4:16pm
Short Link:
Type:
Poster
Event:
Presenter's Name:
Jakub Vít
Paper Code:
4348
Document Year:
2018
Cite

Document Files

poster.pdf

(257 downloads)

Subscribe

[1] Zdeněk Hanzlíček, Jindřich Matoušek, "On the analysis of training data for wavenet-based speech synthesis", IEEE SigPort, 2018. [Online]. Available: http://sigport.org/2759. Accessed: Dec. 18, 2018.
@article{2759-18,
url = {http://sigport.org/2759},
author = {Zdeněk Hanzlíček; Jindřich Matoušek },
publisher = {IEEE SigPort},
title = {On the analysis of training data for wavenet-based speech synthesis},
year = {2018} }
TY - EJOUR
T1 - On the analysis of training data for wavenet-based speech synthesis
AU - Zdeněk Hanzlíček; Jindřich Matoušek
PY - 2018
PB - IEEE SigPort
UR - http://sigport.org/2759
ER -
Zdeněk Hanzlíček, Jindřich Matoušek. (2018). On the analysis of training data for wavenet-based speech synthesis. IEEE SigPort. http://sigport.org/2759
Zdeněk Hanzlíček, Jindřich Matoušek, 2018. On the analysis of training data for wavenet-based speech synthesis. Available at: http://sigport.org/2759.
Zdeněk Hanzlíček, Jindřich Matoušek. (2018). "On the analysis of training data for wavenet-based speech synthesis." Web.
1. Zdeněk Hanzlíček, Jindřich Matoušek. On the analysis of training data for wavenet-based speech synthesis [Internet]. IEEE SigPort; 2018. Available from : http://sigport.org/2759