Neural Generation of Dialogue Response Timings

Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 2442–2452 July 5 - 10, 2020. c⃝2020 Association for Computational Linguistics 2442 Neural Generation of Dialogue Response Timings Matthew Roddy and Naomi Harte ADAPT Centre, School of Engineering Trinity College Dublin, Ireland {roddym,nharte}@tcd.ie Abstract The timings of spoken response offsets in hu- man dialogue have been shown to vary based on contextual elements of the dialogue. We propose neural models that simulate the distri- butions of these response offsets, taking into account the response turn as well as the pre- ceding turn. The models are designed to be integrated into the pipeline of an incremental spoken dialogue system (SDS). We evaluate our models using offline experiments as well as human listening tests. We show that human listeners consider certain response timings to be more natural based on the dialogue context. The introduction of these models into SDS pipelines could increase the perceived natural- ness of interactions.1 1 Introduction The components needed for the design of spoken dialogue systems (SDSs) that can communicate in a realistic human fashion have seen rapid advance- ments in recent years (e.g. Li et al. (2016); Zhou et al. (2018); Skerry-Ryan et al. (2018)). How- ever, an element of natural spoken conversation that is often overlooked in SDS design is the tim- ing of system responses. Many turn-taking com- ponents for SDSs are designed with the objective of avoiding interrupting the user while keeping the lengths of gaps and overlaps as low as possible e.g. Raux and Eskenazi (2009). This approach does not emulate naturalistic response offsets, since in human-human conversation the distributions of re- sponse timing offsets have been shown to differ based on the context of the first speaker’s turn and the context of the addressee’s response (Sacks et al., 1974; Levinson and Torreira, 2015; Heeman and Lunsford, 2017). It has also been shown that lis- teners have different anticipations about upcoming 1 Our code is available at https://github.com/ mattroddy/RTNets. 20000 10000 0 10000 20000 30000 0 100000 200000 300000 30000 20000 10000 0 10000 20000 30000 20000 10000 0 10000 20000 30000 0 100000 200000 300000 30000 20000 10000 0 10000 20000 30000 Response Encoder “Are you doing any kind of volunteer work now?” “Not right now...” Feature extraction Inference LSTM User Turn System Turn 0 0 0 0 0 1 Output Probabilities Sampling 1.5 1.0 0.5 0.0 0.5 1.0 1.5 True Predicted 0 Response Offset (Seconds) Figure 1: Overview of how our model generates the distribution of turn-switch offset timings using an en- coding of a dialogue system response hz, and features extracted from the user’s speech xn. responses based on the length of a silence before a response (B¨ogels et al., 2019). If we wish to real- istically generate offsets distributions in SDSs, we need to design response timing models that take into account the context of the user’s speech and the upcoming system response. For example, off- sets where the first speaker’s turn is a backchannel occur …

Content truncated to 500 words