Speech synthesis technologies for individuals with vocal disabilities: Voice banking and reconstruction

Yamagishi, Junichi; Veaux, Christophe; King, Simon; Renals, Steve

doi:10.1250/ast.33.1

Cited by 86 publications

(56 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…This paper proposes a method to improve naturalness of non-native speech synthesis preserving speaker individuality based on the partial correction of prosodic and phonetic characteristics † , inspired by the previous work on improvements of naturalness of disordered speech for creating a personalized speech synthesis system [24]. The overview of the proposed method is shown in Fig.…”

Section: Figmentioning

confidence: 99%

Non-Native Text-to-Speech Preserving Speaker Individuality Based on Partial Correction of Prosodic and Phonetic Characteristics

Oshima

Takamichi

Toda

et al. 2016

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

SUMMARYThis paper presents a novel non-native speech synthesis technique that preserves the individuality of a non-native speaker. Crosslingual speech synthesis based on voice conversion or Hidden Markov Model (HMM)-based speech synthesis is a technique to synthesize foreign language speech using a target speaker's natural speech uttered in his/her mother tongue. Although the technique holds promise to improve a wide variety of applications, it tends to cause degradation of target speaker's individuality in synthetic speech compared to intra-lingual speech synthesis. This paper proposes a new approach to speech synthesis that preserves speaker individuality by using non-native speech spoken by the target speaker. Although the use of non-native speech makes it possible to preserve the speaker individuality in the synthesized target speech, naturalness is significantly degraded as the synthesized speech waveform is directly affected by unnatural prosody and pronunciation often caused by differences in the linguistic systems of the source and target languages. To improve naturalness while preserving speaker individuality, we propose (1) a prosody correction method based on model adaptation, and (2) a phonetic correction method based on spectrum replacement for unvoiced consonants. The experimental results using English speech uttered by native Japanese speakers demonstrate that (1) the proposed methods are capable of significantly improving naturalness while preserving the speaker individuality in synthetic speech, and (2) the proposed methods also improve intelligibility as confirmed by a dictation test. key words: cross-lingual speech synthesis, English-Read-by-Japanese, speaker individuality, HMM-based speech synthesis, prosody correction, phonetic correction

show abstract

Section: Figmentioning

confidence: 99%

Non-Native Text-to-Speech Preserving Speaker Individuality Based on Partial Correction of Prosodic and Phonetic Characteristics

Oshima

Takamichi

Toda

et al. 2016

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

show abstract

“…MyVoice is CV-mora based, so it is inferior to other software in terms of fluency or clarity [6,7]. However, no other similar software that we know of pursues the second goal as much as MyVoice does.…”

Section: How Myvoice Workmentioning

confidence: 99%

MyVoice: Rescuing voices of ALS patients

Kawahara

Homma

Yoshimura³

et al. 2016

Acoust. Sci. & Tech.

View full text Add to dashboard Cite

“…Veaux et al used HMM-based speech synthesis to reconstruct the voice of individuals with degenerative speech disorders resulting from Amyotrophic Lateral Sclerosis (ALS). Yamagishi et al (Yamagishi et al, 2013) proposed a project named "Voice Banking and Reconstruction". In that project, various types of voices are collected and they proposed TTS for ALS using that database.…”

Section: Related Workmentioning

confidence: 99%

Proceedings of the 5th Workshop on Speech and Language Processing for Assistive Technologies

2014

View full text Add to dashboard Cite

Augmentative Alternative Communication (AAC) policy suffers from a lack of large scale quantitative evidence on the demographics of users and diversity of devices.The 2013 Domesday Dataset was created to aid formation of AAC policy at the national level. The dataset records purchases of AAC technology by the UK's National Health Service between 2006 and 2012; giving information for each item on: make, model, price, year of purchase, and geographic area of purchase. The dataset was designed to help answer open questions about the provision of AAC services in the UK; and the level of detail of the dataset is such that it can be used at the research level to provide context for researchers and to help validate (or not) assumptions about everyday AAC use. This paper examine three different ways of using the Domesday Dataset to provide verified evidence to support, or refute, assumptions, uncover important research problems, and to properly map the technological distinctiveness of a user community. IntroductionTechnical researchers in the AAC community are required to make certain assumptions about the state of the community when choosing research projects that are calculated to make the most effective use of research resources for the greatest possible benefit.A particular issue is estimating how easily technical research can achieve wide scale adoption or commercial impact. For example, (Szekely et al., 2012) uses a webcam and facial analysis to allow a user to control expressive features of their synthetic speech by means of facial expressions. Such work is clearly useful, but it is difficult to assess its potential commercial impact without also knowing what proportion of currently available AAC devices include webcams and how that proportion is changing over time. Similarly, corpus based approaches such as (Mitchell and Sproat, 2012) could potentially be brought to market very quickly, but that potential can only be assessed if we also have some awareness of the range and popularity of AAC devices that either have space for such a corpus or the internet capability to access one. Unfortunately, even though there are a range of AAC focused meta-studies in the literature (see, for example, (Pennington et al., 2003; Pennington et al., 2004; Hanson et al., 2004; Alwell and Cobb, 2009)) they give little information on the technical landscape of AAC. This paper examines three issues of interest to technical researchers in AAC, each from a different stage in the research lifecycle. It then shows how the Domesday Dataset (Reddington, 2013) can provide evidence to support, or refute, assumptions, uncover important research problems, and map the technological distinctiveness of a user community.This paper is structured as follows, Section 2 introduces the Domesday Dataset and discusses the context it is used in in this work. Section 3 examines the issue that little is known about the prevalence of equipment within the AAC user community, and because of this lack of information it is difficult to establish ba...

show abstract

Speech synthesis technologies for individuals with vocal disabilities: Voice banking and reconstruction

Cited by 86 publications

References 15 publications

Non-Native Text-to-Speech Preserving Speaker Individuality Based on Partial Correction of Prosodic and Phonetic Characteristics

Non-Native Text-to-Speech Preserving Speaker Individuality Based on Partial Correction of Prosodic and Phonetic Characteristics

MyVoice: Rescuing voices of ALS patients

Proceedings of the 5th Workshop on Speech and Language Processing for Assistive Technologies

Contact Info

Product

Resources

About