Speakers tend to engage in adaptive behavior, known as entrainment, when they become similar to their interlocutor in various aspects of speaking. We present an unsupervised deep learning framework that derives meaningful representation from textual features for developing semantic entrainment. We investigate the model's performance by extracting features using different variations of the BERT model (DistilBERT and XLM-RoBERTa) and Google's universal sentence encoder (USE) embeddings on two human-human (HH) corpora (The Fisher Corpus English Part 1, Columbia games corpus) and one human-machine (HM) corpus (Voice Assistant Conversation Corpus (VACC)). In addition to semantic features we also trained DNN-based models utilizing two auditory embeddings (TRIpLet Loss network (TRILL) vectors, Low-level descriptors (LLD) features) and two units of analysis (Inter pausal unit and Turn). The results show that semantic entrainment can be assessed with our model, that models can distinguish between HH and HM interactions and that the two units of analysis for extracting acoustic features provide comparable findings.

翻译：说话者在对话中倾向于表现出适应性行为，即在与对话者的交流中，会在多个说话维度上趋于相似，这种现象被称为协同。我们提出了一种无监督深度学习框架，该框架能够从文本特征中提取有意义的表征以构建语义协同模型。通过使用BERT模型的两种变体（DistilBERT与XLM-RoBERTa）以及谷歌通用句子编码器（USE）嵌入，对两个人类-人类（HH）语料库（Fisher英语语料库第一部分、哥伦比亚游戏语料库）和一个人类-机器（HM）语料库（语音助手对话语料库（VACC））进行特征提取，我们分析了模型性能。除语义特征外，我们还基于深度神经网络（DNN）模型，利用两种听觉嵌入（三重态损失网络（TRILL）向量、低级描述符（LLD）特征）和两种分析单元（语间停顿单元与话轮）进行训练。实验结果表明：本模型可有效评估语义协同，模型能够区分人类-人类与人类-机器交互，且用于提取声学特征的两种分析单元获得了可比性的结果。