A Data-Centric Framework for Addressing Phonetic and Prosodic Challenges in Russian Speech Generative Models

Russian speech synthesis presents distinctive challenges, including vowelreduction, consonant devoicing, variable stress patterns, homograph ambiguity,and unnatural intonation. This paper introduces Balalaika, a novel datasetcomprising more than 2,000 hours of studio-quality Russian speech withcomprehensive textual annotations, including punctuation and stress markings.Experimental results show that models trained on Balalaika significantlyoutperform those trained on existing datasets in both speech synthesis andenhancement tasks. We detail the dataset construction pipeline, annotationmethodology, and results of comparative evaluations.