The role of context in neural pitch accent detection in English

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pages 7994–8000, November 16–20, 2020. c⃝2020 Association for Computational Linguistics 7994 The role of context in neural pitch accent detection in English Elizabeth Nielsen Mark Steedman School of InformaticsSharon Goldwater University of Edinburgh, UK e.k.nielsen@sms.ed.ac.uk {steedman, sgwater}@inf.ed.ac.uk Abstract Prosody is a rich information source in natural language, serving as a marker for phenomena such as contrast. In order to make this informa- tion available to downstream tasks, we need a way to detect prosodic events in speech. We propose a new model for pitch accent detec- tion, inspired by the work of Stehwien et al. (2018), who presented a CNN-based model for this task. Our model makes greater use of context by using full utterances as input and adding an LSTM layer. We find that these in- novations lead to an improvement from 87.5 percent to 88.7 percent accuracy on pitch ac- cent detection on American English speech in the Boston University Radio News Corpus, a state-of-the-art result. We also find that a sim- ple baseline that just predicts a pitch accent on every content word yields 82.2 percent accu- racy, and we suggest that this is the appropriate baseline for this task. Finally, we conduct abla- tion tests that show pitch is the most important acoustic feature for this task and this corpus. 1 Introduction Prosody is a rich information source with the poten- tial to improve performance in many spoken NLP tasks (Roesiger et al., 2017; Niemann et al., 1998). In order to make prosodic information available to downstream tasks, many models have been pro- posed to predict which words in an utterance carry pitch accents—word-level prosodic prominences signaled by a deviation from the speaker’s usual pitch, duration, intensity, or some combination of these three features. Identifying pitch accents is helpful since they are often used to signal impor- tant or unexpected information. For example, pitch accents in English typically fall on content words, which are generally more informative. When a pitch accent falls on a function word, it indicates that it is unusually informative, as in the sentence, They ran out of toilet paper even before the quar- antine, where before is more informative because it contrasts with what might be a default assumption (e.g., during). Previous pitch accent prediction models include rule-based models (Brenier et al., 2005), tradi- tional machine learning models (Wightman and Ostendorf, 1994; Levow, 2005; Gregory and Altun, 2004), and neural models (Fernandez et al., 2017; Stehwien and Vu, 2017; Stehwien et al., 2018). Ste- hwien and Vu (2017) and Stehwien et al. (2018) (henceforth, SVS18) showed that neural methods can perform comparably to traditional methods us- ing a relatively small amount of speech context— just a single word on either side of the target word. However, since pitch accents are deviations from a speaker’s average pitch, intensity, and duration, we hypothesize that, as in some non-neural mod- els (e.g. Levow 2005; Rosenberg and Hirschberg 2009), a wider input context will …

Content truncated to 500 words