RO-N3WS: Enhancing Generalization in Low-Resource ASR with Diverse Romanian Speech Benchmarks

TL;DR

RO-N3WS is a diverse Romanian speech dataset that enhances the generalization of low-resource ASR systems, demonstrating significant improvements through fine-tuning and synthetic data augmentation.

Contribution

The paper introduces RO-N3WS, a new benchmark dataset for Romanian speech recognition, and evaluates its effectiveness in improving ASR performance in low-resource and diverse domains.

Findings

Fine-tuning on RO-N3WS significantly reduces WER.

Synthetic TTS data aids in model robustness.

State-of-the-art models benefit from the dataset in low-resource settings.

Abstract

We introduce RO-N3WS, a benchmark Romanian speech dataset designed to improve generalization in automatic speech recognition (ASR), particularly in low-resource and out-of-distribution (OOD) conditions. RO-N3WS comprises over 126 hours of transcribed audio collected from broadcast news, literary audiobooks, film dialogue, children's stories, and conversational podcast speech. This diversity enables robust training and fine-tuning across stylistically distinct domains. We evaluate several state-of-the-art ASR systems (Whisper, Wav2Vec 2.0) in both zero-shot and fine-tuned settings, and conduct controlled comparisons using synthetic data generated with expressive TTS models. Our results show that even limited fine-tuning on real speech from RO-N3WS yields substantial WER improvements over zero-shot baselines. We will release all models, scripts, and data splits to support reproducible research in multilingual ASR, domain adaptation, and lightweight deployment.