When Spoof Detectors Travel: Evaluation Across 66 Languages in the Low-Resource Language Spoofing Corpus
Domain:
natural language processingRecord type:
paperdatasetmodel
Creator:
BorKudMasGor
Host:
We introduce LRLspoof, a large-scale multilingual synthetic-speech corpus for cross-lingual spoof detection, comprising 2,732 hours of audio generated with 24 open-source TTS systems across 66 languages, including 45 low-resource languages under our operational definition. To evaluate robustness without requiring target-domain bonafide speech, we benchmark 11 publicly available countermeasures using threshold transfer: for each model we calibrate an EER operating point on pooled external benchmarks and apply the resulting threshold, reporting spoof rejection rate (SRR). Results show model-dependent cross-lingual disparity, with spoof rejection varying markedly across languages even under controlled conditions, highlighting language as an independent source of domain shift in spoof detection. The dataset is publicly available at \href{huggingface.co and \href{modelscope.cn This paper has been submitted to Interspeech 2026 for review
Similar
Large Language Models as Detectors or Instigators of Hate Speech in Low-resource Ethiopian LanguagesWhen Distributions Shifts: Causal Generalization for Low-Resource LanguagesLow-Resource Corpus Indonesian Local LanguageDo LLM hallucination detectors suffer from low-resource effect?EthioMT: Parallel Corpus for Low-resource Ethiopian LanguagesParaKab – Many Languages, One Kabyle: A Multilingual Parallel Corpus for a Low-Resource Language