Abstract
The rapid growth of speech synthesis and voice conversion systems has made deepfake audio a major security concern. Bengali deepfake detection remains largely unexplored. In this work, we study automatic detection of Bengali audio deepfakes using the BanglaFake dataset. We evaluate zeroshot inference with several pretrained models. These include Wav2Vec2-XLSR-53, Whisper, PANNsCNN14, WavLM and Audio Spectrogram Transformer. Zero-shot results show limited detection ability. The best model, Wav2Vec2-XLSR-53, achieves 53.80% accuracy, 56.60% AUC and 46.20% EER. We then f ine-tune multiple architectures for Bengali deepfake detection. These include Wav2Vec2-Base, LCNN, LCNN-Attention, ResNet18, ViT-B16 and CNN-BiLSTM. Fine-tuned models show strong performance gains. ResNet18 achieves the highest accuracy of 79.17%, F1 score of 79.12%, AUC of 84.37% and EER of 24.35%. Experimental results confirm that fine-tuning significantly improves performance over zero-shot inference. This study provides the first systematic benchmark of Bengali deepfake audio detection. It highlights the effectiveness of f ine-tuned deep learning models for this low-resource language.
Abstract (translated)
语音合成和声音转换系统的迅速发展使得深度伪造音频成为了一个重要的安全问题。然而，孟加拉语的深度伪造检测仍是一个尚未充分探索的领域。在这项工作中，我们使用BanglaFake数据集研究了自动检测孟加拉语深度伪造音频的方法，并评估了几种预训练模型在零样本推理（zeroshot inference）下的性能表现。这些预训练模型包括Wav2Vec2-XLSR-53、Whisper、PANNsCNN14、WavLM和Audio Spectrogram Transformer。零样本推理的结果显示出有限的检测能力，其中最佳模型为Wav2Vec2-XLSR-53，其准确率为53.80%，AUC（Area Under Curve）值为56.60%，EER（等错误率）为46.20%。 接着，我们将多种架构进行了微调以专门用于孟加拉语深度伪造音频的检测。这些模型包括Wav2Vec2-Base、LCNN、LCNN-Attention、ResNet18、ViT-B16和CNN-BiLSTM。经过微调后的模型显示出了显著的性能提升，其中ResNet18取得了最佳的准确率79.17%，F1值为79.12%，AUC值为84.37%以及EER为24.35%。 实验结果证实，相对于零样本推理方法而言，微调策略极大地提高了性能。这项研究提供了孟加拉语深度伪造音频检测的第一个系统性基准测试，并强调了对于这种资源有限的语言来说，经过微调的深度学习模型的有效性。
URL
https://arxiv.org/abs/2512.21702
