Dialectal speech remains largely underexplored in Automatic Speech Recognition (ASR) research, particularly for Slavic languages. While Ukrainian ASR systems have rapidly improved in recent years with the adoption of Whisper, XLS-R, and Wav2Vec-based models, performance on dialectal variants remains unknown and often significantly degraded. In this work, we present the first dedicated effort to build ASR resources for the Hutsul dialect of Ukrainian. We develop a data preparation and segmentation pipeline, evaluate multiple forced alignment strategies, and benchmark state-of-the-art ASR models under zero-shot and fine-tuned conditions. We evaluate results using WER and CER demonstrating that large multilingual ASR models struggle with dialectal speech, while lightweight fine-tuning produces substantial improvements. All scripts, alignment tools, and training recipes are made publicly available to support future research on Ukrainian dialect speech.

We present a significant expansion of ASR resources for the Hutsul dialect of Ukrainian, building on prior work that established the first aligned speech corpus from a single literary source. In this work, we scale the dataset from a single speaker to a multi-speaker corpus comprising 40 speakers and 60.63 hours of audio drawn from diverse sources: YouTube channels (with author permissions), field recordings from native speakers, linguist student recordings, and regional radio broadcasts. To obtain reference transcriptions for audio without existing text, we introduce a novel RAG-enhanced correction pipeline: audio is first transcribed using ElevenLabs, then corrected through a RAG pipeline backed by a dialect-aware language model. We evaluate a fine-tuned ASR models across five distinct speaker datasets, demonstrating that while the model achieves strong performance on in-domain speakers (CER 3.24%), cross-speaker generalization remains challenging, with CER ranging from 5.33% to 17.24% depending on speaker characteristics. All data, code, and models are released publicly to support further research on Ukrainian dialect speech technologies.

Adapting large language models to low-resource languages presents three interconnected challenges: inefficient tokenization, scarcity of high-quality annotated data, and limited resources for instruction tuning. We present a reproducible approach that addresses each challenge using data-centric methods that primarily rely on unlabeled text corpora, parallel translation data, and a multilingual base model. Our approach combines (1) vocabulary surgery for tokenizer adaptation without full retraining, (2) cross-lingual transfer of quality classifiers via translation, enabling filtering without target-language annotations, and (3) generation of instruction data through translation, task conversion, and targeted synthesis. We validate this recipe by adapting Gemma-3-12B to Ukrainian. %, producing Lapa-12BOur pretrained model achieves top performance on Ukrainian benchmarks, while our instruction-tuned variant demonstrates strong performance on translation (33 BLEU on FLORES), summarization, and question-answering tasks, while requiring 1.5x fewer tokens than the original model for the same text. We release all models, datasets, classifiers, and code to enable replication for other languages.