Modeling Intra-word Code-Switching for Karelian ASR. / Киселева, Ксения Олеговна; Кипяткова, Ирина Сергеевна; Кагиров, Ильдар Амирович; Долгушин, Михаил Дмитриевич.
Speech and Computer 27th International Conference, SPECOM 2025, Szeged, Hungary, October 13–15, 2025, Proceedings, Part II. Springer Nature, 2026. p. 104-117 (Lecture Notes in Artificial Intelligence; Vol. 16188).Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Research › peer-review
}
TY - GEN
T1 - Modeling Intra-word Code-Switching for Karelian ASR
AU - Киселева, Ксения Олеговна
AU - Кипяткова, Ирина Сергеевна
AU - Кагиров, Ильдар Амирович
AU - Долгушин, Михаил Дмитриевич
N1 - Conference code: 27
PY - 2026
Y1 - 2026
N2 - This paper addresses issues of modeling Karelian-Russian code-switching for automatic speech recognition, with a focus on intra-word code-switching. Due to grammatical differences between Karelian and Russian, and the lack of automatic translation tools for languages in question, standard augmentation methods relying on parallel translated text corpora are difficult to apply. To address these issues, we developed a set of rules specifically designed for generating words with intra-word code-switching, and then augmented the Karelian text by substituting random words with their corresponding generated counterparts. Besides that, we performed linear interpolation of the Karelian language model with the Russian one. We fine-tuned Wav2Vec2.0-large-uralic-voxpopuli-v2 on both Karelian and Russian speech data with the further integration of the developed language model into the system. An evaluation demonstrates significant accuracy improvement: compared to the baseline system without a language model, we achieved relative WER reductions of 11.3% on the development set and 16.6% on the test set.
AB - This paper addresses issues of modeling Karelian-Russian code-switching for automatic speech recognition, with a focus on intra-word code-switching. Due to grammatical differences between Karelian and Russian, and the lack of automatic translation tools for languages in question, standard augmentation methods relying on parallel translated text corpora are difficult to apply. To address these issues, we developed a set of rules specifically designed for generating words with intra-word code-switching, and then augmented the Karelian text by substituting random words with their corresponding generated counterparts. Besides that, we performed linear interpolation of the Karelian language model with the Russian one. We fine-tuned Wav2Vec2.0-large-uralic-voxpopuli-v2 on both Karelian and Russian speech data with the further integration of the developed language model into the system. An evaluation demonstrates significant accuracy improvement: compared to the baseline system without a language model, we achieved relative WER reductions of 11.3% on the development set and 16.6% on the test set.
KW - Automatic Speech Recognition
KW - Code-Switching
KW - Language Modeling
KW - Livvi-Karelian
UR - https://www.mendeley.com/catalogue/ca6153f9-0d3b-376e-9117-dae2d7b8d8c4/
U2 - 10.1007/978-3-032-07959-6_8
DO - 10.1007/978-3-032-07959-6_8
M3 - Conference contribution
SN - 9783032079589
T3 - Lecture Notes in Artificial Intelligence
SP - 104
EP - 117
BT - Speech and Computer 27th International Conference, SPECOM 2025, Szeged, Hungary, October 13–15, 2025, Proceedings, Part II
PB - Springer Nature
T2 - 27th International Conference on Speech and Computer
Y2 - 13 October 2025 through 14 October 2025
ER -
ID: 142747829