Vocoder+DNN submission

Details of the submission

Mohammed Salah Al-Radhi, Tamás Gábor Csapó, Géza Németh, ,,Noise and acoustic modeling with waveform generator in text-to-speech and neutral speech conversion'', submitted.

Samples of the listening test #1

sentenceNaturalAnchorHMM
(Continuous vocoder)		DNN
(WORLD vocoder)		DNN
(Continuous vocoder)		DNN
(Continuous vocoder
+ envelope)
SLT / 001
SLT / 002
SLT / 003
SLT / 004
SLT / 005
SLT / 006
SLT / 007
SLT / 008
SLT / 009
SLT / 010
SLT / 011
SLT / 012
SLT / 013
SLT / 014
SLT / 015

Samples of the listening test #2

sentenceNaturalAnchorDNNBLSTMHybrid
AWB / 01
AWB / 02
AWB / 03
AWB / 04
AWB / 05
AWB / 06
AWB / 07
AWB / 08
AWB / 09
AWB / 10
SLT / 01
SLT / 02
SLT / 03
SLT / 04
SLT / 05
SLT / 06
SLT / 07
SLT / 08
SLT / 09
SLT / 10

Samples of the listening test #3 (VC)

sentenceTargetSourceVoice Conversion
(Sprocket)		Voice Conversion
(MagPhase)		Voice Conversion
(WORLD)		Voice Conversion
(CSM - Proposed)
BDL-to-CLB
BDL-to-JMK
BDL-to-SLT
CLB-to-BDL
CLB-to-JMK
CLB-to-SLT
JMK-to-BDL
JMK-to-CLB
JMK-to-SLT
SLT-to-BDL
SLT-to-CLB
SLT-to-JMK


Our laboratory deals with the development of Hungarian and foreign language text-to-speech solutions, investigation of human-computer interactions and the research of modern machine learning algorithms.

Our academic and industrial partners are successfully applying our solutions.

Featured contents

For students

Contact

Magyar tudósok krt. 2.
1117 Budapest, HUNGARY
Phone: +36-1-463-3883
Fax: +36-1-463-3107
Email: smartlab@tmit.bme.hu

2024 © All rights reserved. Except as permitted by the copyright law applicable to you, you may not reproduce or communicate any of the content on this website, including files downloadable from this website, without the permission of the copyright owner. (v914)