Kazlauskienė, Asta

The aim of this research is to determine whether pauses, decreases in intensity, pitch lowering, lengthening of the final sound, and certain voice quality features are equally significant in marking phrase boundaries at different reading rates. The research material consists of 9 recordings of Aesop's fable "The North Wind and the Sun". This text was read by three male actors 3 times: at a natural reading rate, at a fast reading rate, and at a slow reading rate. The presence and duration of pauses, the intensity and F0 of the entire phrase and its final syllable, as well as the duration, intensity, F0, jitter, shimmer, and HNR of the final vowel at the end and in phrase-internal position, were analysed. The results indicate that all parameters can serve as indicators of phrase boundaries, particularly for intonational phrases. Intonational phrases can be differentiated from intermediate phrases based on the level of intensity decrease, jitter, and to some extent, pauses and F0 lowering. However, among these parameters, only duration-related features, such as pauses and the lengthening of the final sound, show a more prominent sensitivity to changes in reading rate.

This study investigates whether specific acoustic features can reliably indicate phrase boundaries for automatic detection. It includes (1) an analysis of acoustic markers at the end of prosodic units—intonational phrases, intermediate phrases, and words—and (2) the evaluation of these features in an automatic boundary detection algorithm. Data were drawn from professionally and expressively read speech (893 words), news broadcasts (732 words), and interviews (361 words). Key features analyzed were pause duration, final sound lengthening, intensity, and F0 changes. Findings show that pauses and their duration are the most consistent indicators of phrase boundaries, especially at intonational phrase ends. Final sound lengthening and reductions in intensity and F0 also contribute but are less reliable for intermediate phrases. In automatic detection phonetic cues can be used to predict boundaries assigned by phoneticians 69% of the time. Read speech yielded better results than spontaneous speech. Among the features, pause presence and length were the most reliable, while F0 and intensity changes played a minor role

The lack of robust and accessible speech resources for the Lithuanian language poses a challenge to its digitization, including efforts related to speech recognition, artificial intelligence (AI), and language technologies. This issue is acknowledged in the State Digitization Development Program 2021-2030 of the Ministry of the Economy and Innovation of the Republic of Lithuania. The program emphasizes the need to integrate advanced tools and technological solutions to improve the accessibility, security, and efficiency of e-services at both the national and international levels. The project “Development of the Large Lithuanian Speech Corpus (LIEPA-3)” contributes to the digitization of the Lithuanian language. During the project, a large corpus of 10,000 hours of annotated speech will be created. The sources will consist of 5000 hours of read speech, 4900 hours of spontaneous speech and 100 hours of four Lithuanian dialects. The corpus will be freely available in open formats on different platforms. The results of this project will facilitate innovation in AI and digital services, as well as improve public access to e-services in Lithuania. It will simplify interactions with digital platforms, promote digital inclusion and contribute to the broader adoption of AI technologies in various sectors. Ultimately, LIEPA-3 will support a more connected and digitally literate society by providing essential resources for developing userfriendly digital services. Also, the created speech corpus will be a good source for researchers in many fields, primarily contributing to linguistic research and informatics.

In many languages, pitch changes at phrase edges are often described as significant cues of juncture. The aim of this study is to analyse pitch changes at the beginning and end of declarative phrases in standard Lithuanian. The study consists of two parts: 1) a pilot study analysing whether the F0 difference is a reliable measure to determine the F0 rise/fall at the end of declarative phrasesand whether such a difference depends on the speaker’s sex; 2) an annotated speech corpus analysis on the tendencies of pitch changes at phrase edges. The pilot study showed that in phrases with a rising F0 at the end, the F0 mean of the last stressed syllable is significantly lower than F0 at the end of a phrase. In phrases with a falling F0, the F0 mean of the last stressed syllable is significantly higher than the F0 at the end of a phrase. The F0 difference is a reliable measure to determine the F0 rise or fall at the end of a phrase, and all the abovementioned tendencies were found in both male and female speakers’ recordings. The annotated speech corpus analysis revealed that both intonational and intermediate phrases usually begin and end in a low tone. Hence, the high and rising or falling tones at the beginning or end of a phrase are the marked members of the phrase edge tone system, and they can help identify phrase junctures. However, it is important to pay attention to other acoustic features such as pauses and intensity, as well as grammatical and semantic aspects

Monografijoje: 1) pateikiama trumpa artikuliacinių ir akustinių intonacijos koreliatų, intonacijos tyrimų apžvalga, preliminarios įžvalgos dėl galimos lietuvių kalbos prozodinės hierarchijos; 2) analizuojamos frazės centro – loginį arba frazės kirtį turinčio žodžio – akustinės ypatybės; 3) gilinamasi į akustines ir psichoakustines frazių sandūrų ir pakraščių ypatybes, jų suvokimo dėsningumus; 4) aiškinamasi intonacijos žymėjimo specialiems poreikiams ir įprastame rašytiniame tekste ypatumai, pateikiamos rekomendacijos; 5) nagrinėjamas intonacijos ir skyrybos santykis rašytinėje kalboje.

Monografijoje aprašomo tyrimo tikslas – nustatyti ir aprašyti bendrinės lietuvių kalbos skiemens ypatybes, jo struktūros dėsningumus ir skiemenavimą. Tiriamąją medžiagą sudaro daugiau nei 311 tūkst. skiemenų (109 tūkst. žodžių). Lietuvių kalbos skiemuo nagrinėjamas ir aprašomas remiantis įvairiomis teorinėmis prieigomis: dėmenų teorijos požiūriu, skiemens svorio ir skiemenų foneminės struktūros (struktūrinių ir distribucinių modelių) nustatymo principais. Remiantis garsų sonoringumo ir optimalumo teorijomis bandoma rasti skiemenų ribas ir išsiaiškinti kalbos vartotojų skiemenavimo polinkius.