IIITH-BUT system for IWSLT 2025 low-resource Bhojpuri to Hindi speech translation

TL;DR

This paper details the IIITH-BUT system's approach to improving low-resource Bhojpuri-Hindi speech translation by hyperparameter tuning, data augmentation, and cross-lingual training, resulting in significant quality enhancements.

Contribution

The paper introduces systematic hyperparameter optimization, data augmentation, and cross-lingual training techniques tailored for low-resource speech translation tasks.

Findings

Hyperparameter tuning improves translation quality.

Data augmentation techniques like speed perturbation and SpecAugment are effective.

Cross-lingual training with Marathi enhances performance.

Abstract

This paper presents the submission of IIITH-BUT to the IWSLT 2025 shared task on speech translation for the low-resource Bhojpuri-Hindi language pair. We explored the impact of hyperparameter optimisation and data augmentation techniques on the performance of the SeamlessM4T model fine-tuned for this specific task. We systematically investigated a range of hyperparameters including learning rate schedules, number of update steps, warm-up steps, label smoothing, and batch sizes; and report their effect on translation quality. To address data scarcity, we applied speed perturbation and SpecAugment and studied their effect on translation quality. We also examined the use of cross-lingual signal through joint training with Marathi and Bhojpuri speech data. Our experiments reveal that careful selection of hyperparameters and the application of simple yet effective augmentation techniques significantly improve performance in low-resource settings. We also analysed the translation hypotheses to understand various kinds of errors that impacted the translation quality in terms of BLEU.