| system: | |
| name: "RawTFNet" | |
| slug: "rawtfnet" | |
| description: > | |
| Lightweight raw-waveform CNN for speech anti-spoofing: sinc-convolution | |
| front-end, depthwise-separable Res2Net-SE conv blocks, and a Tf-SepNet | |
| (time-frequency separable) classifier (depth=10, width=32). ASVspoof2019 LA | |
| pretrained, FP32, deterministic first-64600-sample window (no random crop). | |
| code: "https://github.com/swagshaw/RawTFNet-Pytorch" | |
| checkpoint: "https://huggingface.co/SpeechAntiSpoofingBenchmarks/RawTFNet/blob/aa12f0fe2f10cc5278c954175a12c18cc43e3113/Best_RawTFNet_32.pth" | |
| params_millions: 0.177540 | |
| paper: | |
| arxiv_id: "2507.08227" | |
| url: "https://arxiv.org/abs/2507.08227" | |
| bibtex: | | |
| @article{xiao2025rawtfnet, | |
| title={RawTFNet: A Lightweight CNN Architecture for Speech Anti-spoofing}, | |
| author={Xiao, Yang and Dang, Ting and Das, Rohan Kumar}, | |
| journal={arXiv preprint arXiv:2507.08227}, | |
| year={2025} | |
| } | |
| notes: "Deterministic first-64600-sample window (no random crop). Full RawTFNet (width=32), Best_RawTFNet_32.pth." | |