Re-enacted and spontaneous conversational prosody: How different?

Abstract

Previous work has shown that read and spontaneous monologues differ prosodically both in production and perception. In this paper, we examine whether similar effects can be found between spontaneous and read, or rather acted, dialogues. It is possible that speakers can mimic conversational prosody very well. Alternatively, they might use prosodic resources more than the conversational situation actually requires (overacting). Another possibility is that in acted dialogues, prosody is actually used less as a communicative device, as there is no need to establish a common ground or to organize the floor between interlocutors. In our study, we examined spontaneous and read dialogues of equal verbal content. The task-oriented dialogues contained a communicative situation implicitly demanding for for a higher speaking rate (time pressure). Our results show that globally, speakers met this conversational demand of increased speaking rate both in the acted and in the spontaneous situation, although we find different global speaking rates between the spontaneous and the acted condition. Also, read speech exhibits a lower F0 minimum and, consequently, a larger F0 range than read speech, which may be explicable by a lack of active turn taking organization. Summing up, acted conversational prosody resembles many features of spontaneous interaction, but also shows systematic differences.

Virtual reality simulations as a new tool for practicing presentations and refining public-speaking skills

“…In accord with our prosodic expectations, we found evidence that the speakers of the VR test group gave their elevator pitches in a more listener-oriented, dialogue speaking style than the speakers of the control group. Compared to the latter group, the presentations of the VR test group were characterized by a higher F0 level, a larger F0 range, and a slower speaking rate [6,9,12,13,14,15,16]. Moreover, matching the relatively large speaker-listener distance in the VR situation, the VR test group speakers gave their elevator pitches at a higher intensity level [17,18].…”

“…Although read speech and spontaneous speech both constitute a spectrum of styles depending on medium, preparation, and the amount of interaction [7,8,9,10,11], there 9th International Conference on Speech Prosody 2018 13-16 June 2018, Poznań, Poland are recurring acoustic features associated with the differences between spontaneous and read speech. Compared to readspeech monologues, spontaneous speech dialogues show, for example, a lower speaking rate [5,12,13,14], a higher F0 mean [9] and a larger F0 variability [6,9,12,15,16].…”

God as Interlocutor — Real or Imaginary? Prosodic Markers of Dialogue Speech and Expected Efficacy in Spoken Prayer

“…Yet other studies present differences between read and spontaneous dialogues (e.g., [9]), which we can then relate to contrastive analyses of read and spontaneous monologues [10,11,12]. The overall results patterns are not fully consistent across studies [13]. However, what crystallizes across studies is that, compared to spoken monologues, dialogue speech is characterized by a slower speaking rate as well as by a higher level and a larger range of the fundamental frequency or f0 (i.e.…”

“…the lowest frequency of the produced speech signal that provides the acoustic basis for listeners perception of pitch and, thus, speech melody, [14]). Studies by [15] and [16], amongst others, also found that dialogues show significantly more silent pauses and hence shorter prosodic phrase durations than monologues ( [17], see also [13]), or faked dialogues (without any addressee).…”

Strategies for developing a Conversational Speech Dataset for Text-To-Speech Synthesis

“…We set the floor and ceiling frequency values to 75Hz and 400Hz respectively. Utilizing an approach implemented in [16] we filtered out frequency values that were outside 1.5 times the interquantile range for the speaker. The extracted intonational features include F0mean, F0std, and F0range and are plotted in Figure 3.…”