Saved in:
| Main Authors: | , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2505.13404 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916749047234560 |
|---|---|
| author | Koluguri, Nithin Rao Sekoyan, Monica Zelenfroynd, George Meister, Sasha Ding, Shuoyang Kostandian, Sofia Huang, He Karpov, Nikolay Balam, Jagadeesh Lavrukhin, Vitaly Peng, Yifan Papi, Sara Gaido, Marco Brutti, Alessio Ginsburg, Boris |
| author_facet | Koluguri, Nithin Rao Sekoyan, Monica Zelenfroynd, George Meister, Sasha Ding, Shuoyang Kostandian, Sofia Huang, He Karpov, Nikolay Balam, Jagadeesh Lavrukhin, Vitaly Peng, Yifan Papi, Sara Gaido, Marco Brutti, Alessio Ginsburg, Boris |
| contents | Multi-task and multilingual approaches benefit large models, yet speech processing for low-resource languages remains underexplored due to data scarcity. To address this, we present Granary, a large-scale collection of speech datasets for recognition and translation across 25 European languages. This is the first open-source effort at this scale for both transcription and translation. We enhance data quality using a pseudo-labeling pipeline with segmentation, two-pass inference, hallucination filtering, and punctuation restoration. We further generate translation pairs from pseudo-labeled transcriptions using EuroLLM, followed by a data filtration pipeline. Designed for efficiency, our pipeline processes vast amount of data within hours. We assess models trained on processed data by comparing their performance on previously curated datasets for both high- and low-resource languages. Our findings show that these models achieve similar performance using approx. 50% less data. Dataset will be made available at https://hf.co/datasets/nvidia/Granary |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_13404 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Granary: Speech Recognition and Translation Dataset in 25 European Languages Koluguri, Nithin Rao Sekoyan, Monica Zelenfroynd, George Meister, Sasha Ding, Shuoyang Kostandian, Sofia Huang, He Karpov, Nikolay Balam, Jagadeesh Lavrukhin, Vitaly Peng, Yifan Papi, Sara Gaido, Marco Brutti, Alessio Ginsburg, Boris Computation and Language Audio and Speech Processing Multi-task and multilingual approaches benefit large models, yet speech processing for low-resource languages remains underexplored due to data scarcity. To address this, we present Granary, a large-scale collection of speech datasets for recognition and translation across 25 European languages. This is the first open-source effort at this scale for both transcription and translation. We enhance data quality using a pseudo-labeling pipeline with segmentation, two-pass inference, hallucination filtering, and punctuation restoration. We further generate translation pairs from pseudo-labeled transcriptions using EuroLLM, followed by a data filtration pipeline. Designed for efficiency, our pipeline processes vast amount of data within hours. We assess models trained on processed data by comparing their performance on previously curated datasets for both high- and low-resource languages. Our findings show that these models achieve similar performance using approx. 50% less data. Dataset will be made available at https://hf.co/datasets/nvidia/Granary |
| title | Granary: Speech Recognition and Translation Dataset in 25 European Languages |
| topic | Computation and Language Audio and Speech Processing |
| url | https://arxiv.org/abs/2505.13404 |