Granary: Speech Recognition and Translation Dataset in 25 European Languages
This addresses data scarcity for speech processing in low-resource European languages, though it is incremental as it builds on existing pseudo-labeling and data processing techniques.
The authors tackled the problem of data scarcity for speech processing in low-resource languages by creating Granary, a large-scale open-source dataset for speech recognition and translation across 25 European languages, and found that models trained on their processed data achieved similar performance using approximately 50% less data.
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