Naturalness

About: Naturalness is a research topic. Over the lifetime, 1305 publications have been published within this topic receiving 31737 citations.

Speech naturalness ratings and perceptual correlates of highly natural and unnatural speech in hypokinetic dysarthria secondary to Parkinson’s disease

Abstract: 1. IntroductionIn the field of speech-language pathology, speech naturalness is a useful measure of how an individual's participation at the society level1 may be affected by a speech disorder. Depending on the social, and possibly professional, roles the individual wishes to maintain, the clinician will prioritize speech naturalness accordingly when planning intervention. A common emphasis in speech therapy is increasing the intelligibility of speech, keeping with the tradition of functional communicative goals. However, an increase in intelligibility may be accompanied by a decrease in how natural the individual's speech sounds and vice versa (Yorkston, Beukelman, Strand, and Bell, 1999). Therefore, a common dilemma experienced by clinicians is how to balance intelligibility and naturalness.Due to the high level of motor control required for 'normal' or natural speech, clinicians' therapy goals for dysarthria clients are for 'best possible speech' and may not meet the clients' own expectations for their speech after therapy (Yorkston et al., 1999). For the client, sounding 'natural' may be speaking like she or he did before the onset of dysarthria, even if such a goal is out of reach. Consequently, speech-language pathologists must evaluate functional goals in respect to clients' goals. For example, a speaker whose profession requires public speaking will be very concerned with maintaining speech naturalness, but will also want to remain intelligible. Furthermore, from studies of how people are perceived based on their 'tone' of voice, it is understandable why clients would want to speak as naturally as possible, as personal and professional interactions can be affected by listener perceptions of how a person sounds when he or she speaks. For example, a study of doctorpatient interactions found that doctors perceived to have a 'dominant' tone of voice were more likely to be sued for malpractice than those judged to express concern or anxiety when talking with patients (Ambady et al., 2002). Also, as illustrated by the work of Miller and colleagues, the effect of a speech disorder such as dysarthria is felt in social interactions beyond the workplace (Miller et al., 2006, 2008). The potential for a disconnect between clinician and client expectations, along with limited ability to provide a prognosis of speech naturalness based on speech characteristics, can lead to lack of client motivation and compliance with therapy.For that reason, clinicians are paying increasing attention to naturalness in the clinic, but face several limitations with the current state of research on speech naturalness. While there are informal assessment procedures in use for the naturalness of dysarthric speech, no standardized assessment or evidence of what type of speech sample for assessment is most suitable exists. Finally, because there is limited research on the relationship between various speech characteristics and naturalness, current therapeutic approaches that attempt to maximize both the intelligibility and naturalness of a client's speech must proceed on a trial-and-error basis (Yorkston et al., 1999).Unfortunately for the speech-language pathologist consulting research on naturalness and speech disorders, many questions one might ask when attempting to balance these goals remain largely unanswered. The concept of speech naturalness is commonly found in assessment and research, yet it is often left undefined or inconsistently defined. This inconsistency poses a problem for researchers and speech-language pathologists attempting to maintain consistency between procedures and assessments. One reason for this lack of consistency is the use of the terms 'bizarreness', 'normalcy', 'acceptability', 'severity', 'proficiency', and even 'intelligibility' interchangeably with the concept of naturalness (Whitehill, 2002). Another difficulty in defining the concept may lie in the phenomenon of easily recognizing and identifying something that is seen or heard, but performing significantly worse when asked to describe the same thing, a tendency known as verbal overshadowing (Dodson, Johnson, and Schooler, 1997; Schooler and Engstler-Schooler, 1990). …

Dubbing in Practice: A Large Scale Study of Human Localization With Insights for Automatic Dubbing

Abstract: Abstract We investigate how humans perform the task of dubbing video content from one language into another, leveraging a novel corpus of 319.57 hours of video from 54 professionally produced titles. This is the first such large-scale study we are aware of. The results challenge a number of assumptions commonly made in both qualitative literature on human dubbing and machine-learning literature on automatic dubbing, arguing for the importance of vocal naturalness and translation quality over commonly emphasized isometric (character length) and lip-sync constraints, and for a more qualified view of the importance of isochronic (timing) constraints. We also find substantial influence of the source-side audio on human dubs through channels other than the words of the translation, pointing to the need for research on ways to preserve speech characteristics, as well as transfer of semantic properties such as emphasis and emotion, in automatic dubbing systems.

Erratum: Off-Shell Higgs Probe of Naturalness [Phys. Rev. Lett. 120, 111801 (2018)].

Abstract: This corrects the article DOI: 10.1103/PhysRevLett.120.111801.

Naturalness Enhancement with Linguistic Information in End-to-End TTS Using Unsupervised Parallel Encoding.

Abstract: State-of-the-art end-to-end speech synthesis models have reached levels of quality close to human capabilities. However, there is still room for improvement in terms of naturalness, related to prosody, which is essential for human-machine interaction. Therefore, part of current research has shift its focus on improving this aspect with many solutions, which mainly involve prosody adaptability or control. In this work, we explored a way to include linguistic features into the sequenceto-sequence Tacotron2 system to improve the naturalness of the generated voice. That is, making the prosody of the synthesis looking more like the real human speaker. Specifically we embedded with an additional encoder part-of-speech tags and punctuation mark locations of the input text to condition Tacotron2 generation. We propose two different architectures for this parallel encoder: one based on a stack of convolutional plus recurrent layers, and another formed by a stack of bidirectional recurrent plus linear layers. To evaluate the similarity between real read-speech and synthesis, we carried out an objective test using signal processing metrics and a perceptual test. The presented results show that we achieved an improvement in naturalness.

What Makes an Explanation

Abstract: Newtonian theory has usually been accepted as a paradigm example of an explanation. There are two widely known analyses of what makes it so. According to one analysis, the deductive and predictive nature of the theory is what counts. The second analysis emphasizes the ability of the theory to connect widely different events and laws. The present paper proposes a third analysis stressing three characteristics. (1) The explanation includes a description which is in part of something unobserved. (2) The description is true in the sense of corresponding to the facts. (3) Through the description, the explanation confers "naturalness" upon the thing explained.