CoRuSS - a New Prosodically Annotated Corpus of Russian Spontaneous Speech

CoRuSS — a New Prosodically Annotated Corpus of Russian Spontaneous Speech Tatiana Kachkovskaia, Daniil Kocharov, Pavel Skrelin, Nina Volskaya Department of Phonetics, Saint Petersburg State University, St.Petersburg State University, 7/9 Universitetskaya nab., St. Petersburg, 199034 Russia [kachkovskaia, kocharov, skrelin, volni]@phonetics.pu.ru Abstract This paper describes speech data recording, processing and annotation of a new speech corpus CoRuSS (Corpus of Russian Spontaneous Speech), which is based on connected communicative speech recorded from 60 native Russian male and female speakers of different age groups (from 16 to 77). Some Russian speech corpora available at the moment contain plain orthographic texts and provide some kind of limited annotation, but there are no corpora providing detailed prosodic annotation of spontaneous conversational speech. This corpus contains 30 hours of high quality recorded spontaneous Russian speech, half of it has been transcribed and prosodically labeled. The recordings consist of dialogues between two speakers, monologues (speakers’ self-presentations) and reading of a short phonetically balanced text. Since the corpus is labeled for a wide range of linguistic—phonetic and prosodic—information, it provides basis for empirical studies of various spontaneous speech phenomena as well as for comparison with those we observe in prepared read speech. Since the corpus is designed as a open-access resource of speech data, it will also make possible to advance corpus-based analysis of spontaneous speech data across languages and speech technology development as well. Keywords: speech corpus, speech annotation, Russian 1. Introduction At the moment a number of medium and large size Russian speech corpora are available. The largest published corpus of Russian speech is ORD (One Day of Speech) corpus that is still under development (Bogdanova-Beglarian et al., 2015). It contains more than 1000 hours of everyday speech. The corpus is collected by recording the whole day speech of more than 100 partici- pants. It has partial annotation and transcription. However, this corpus is not publicly available. The most annotated publicly available corpus nowadays is PrACS-Russ (Prosodically Annotated Corpus of Spo- ken Russian), which contains over 4 hours of mono- logue speech (Kibrik and Podlesskaya, 2009) with limited prosodic annotation based on two types of tones—rise and fall—and their combinations. The annotation also includes a detailed labelling of various speech phenomena such as speech errors, self-corrections, filled pauses. It is avail- able as part of Russian National Corpus (Apresjan et al., 2006). The corpus may be a valuable source for research of discourse events in monologue speech, but rather poor recording quality makes it possible to use it only for basic phonetic research. Due to inconsistent and variable noise, these recordings are not suitable for acoustic analysis and modelling in the field of speech technologies. The corpora containing well-annotated high-quality record- ings are not publicly available. It is worth mentioning some of them for their well-structured design and annotation. Corpus of Professionally Read Speech (CORPRES) con- tains over 30 hours of speech recorded in a professional studio (Skrelin et al., 2010). Segmented and annotated on orthographic, phonetic, and prosodic levels, it contains a manually corrected pitch …

Content truncated to 500 words