Rhythm analysis in Arabic L2 speech

Abstract

In this paper we present evidence that speech produced spontaneously in a conversation is considered more natural than read prompts. We also explore the relationship between participants' expectations of the speech style under evaluation and their actual ratings. In successive listening tests subjects rated the naturalness of either spontaneously produced, read aloud or written sentences, with instructions toward either conversational, reading or general naturalness. It was found that, when presented with spontaneous or read aloud speech, participants consistently rated spontaneous speech more natural-even when asked to rate naturalness in the reading case. Presented with only text, participants generally preferred transcriptions of spontaneous utterances, except when asked to evaluate naturalness in terms of reading aloud. This has implications for the application of MOS-scale naturalness ratings in Speech Synthesis, and potentially on the type of data suitable for use both in general TTS, dialogue systems and specifically in Conversational TTS, in which the goal is to reproduce speech as it is produced in a spontaneous conversational setting.

Testing the consistency assumption: Pronunciation variant forced alignment in read and spontaneous speech synthesis

“…Natural speech is, unsurprisingly, rated significantly higher than synthetic speech. In contrast to [13] there was no significant difference between the natural speech types. However, this is probably due to the natural speech being so clearly more natural than the synthetic, causing differences between these two types to become much smaller.…”

“…In order to support the claim that spontaneous speech is less conformative to standard phonemisations than read speech, and to obtain a gold standard development set, the small corpus of read and spontaneous sentences from [13] was analysed. The corpus contains 50 sentences which were uttered in a normal conversation by a British English female voice talent.…”

Speech Synthesis Evaluation — State-of-the-Art Assessment and Suggestion for a Novel Research Program

“…Still, this remains an underlying assumption in much TTS evaluation research, where this reference or gold standard is often taken as being equivalent to "human read speech". Some researchers criticize this implicit assumption, and postulate an alternative reference such as "conversational speech" [19]. While such an approach may be useful for a particular application such as dialogue systems research, neither speaking style is inherently "neutral" or "natural": Read speech is entirely appropriate in certain contexts of human communication, e.g.…”

“…We therefore contend that even if the goal of the TTS evaluation is a "pure" system comparison, without an actual application in mind, some kind of conceptual framing may be advisable. Indeed, [19] report that simply asking listeners to imagine a particular interactive situation, had an effect on listener's impressions. If no such framing is provided, listeners are forced to imagine some context in which they may listen to the TTS, and are prone to come up with a corresponding set of quality dimensions.…”