Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.798899
Title: Prosody generation for text-to-speech synthesis
Author: Ronanki, Srikanth
ISNI:       0000 0004 8508 9728
Awarding Body: University of Edinburgh
Current Institution: University of Edinburgh
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Abstract:
The absence of convincing intonation makes current parametric speech synthesis systems sound dull and lifeless, even when trained on expressive speech data. Typically, these systems use regression techniques to predict the fundamental frequency (F0) frame-by-frame. This approach leads to overlysmooth pitch contours and fails to construct an appropriate prosodic structure across the full utterance. In order to capture and reproduce larger-scale pitch patterns, we propose a template-based approach for automatic F0 generation, where per-syllable pitch-contour templates (from a small, automatically learned set) are predicted by a recurrent neural network (RNN). The use of syllable templates mitigates the over-smoothing problem and is able to reproduce pitch patterns observed in the data. The use of an RNN, paired with connectionist temporal classification (CTC), enables the prediction of structure in the pitch contour spanning the entire utterance. This novel F0 prediction system is used alongside separate LSTMs for predicting phone durations and the other acoustic features, to construct a complete text-to-speech system. Later, we investigate the benefits of including long-range dependencies in duration prediction at frame-level using uni-directional recurrent neural networks. Since prosody is a supra-segmental property, we consider an alternate approach to intonation generation which exploits long-term dependencies of F0 by effective modelling of linguistic features using recurrent neural networks. For this purpose, we propose a hierarchical encoder-decoder and multi-resolution parallel encoder where the encoder takes word and higher level linguistic features at the input and upsamples them to phone-level through a series of hidden layers and is integrated into a Hybrid system which is then submitted to Blizzard challenge workshop. We then highlight some of the issues in current approaches and a plan for future directions of investigation is outlined along with on-going work.
Supervisor: King, Simon ; Renals, Stephen Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.798899  DOI: Not available
Keywords: speech synthesis ; prosody
Share: