Datasets:
YAML Metadata Error:Invalid content in eval.yaml.
Check out the documentation for more information.
Show details
✖ Invalid input
Download eval.yaml from shanthini33/ASVspoof5: direct link, hf CLI and curl.
- Browser
- Download file 534 Bytes
-
https://huggingface.co/datasets/shanthini33/ASVspoof5/resolve/main/eval.yaml
- Command line
-
hf download hf://datasets/shanthini33/ASVspoof5/eval.yaml
-
curl -L -o eval.yaml https://huggingface.co/datasets/shanthini33/ASVspoof5/resolve/main/eval.yaml
534 Bytes
| name: ASVspoof5 | |
| description: > | |
| Track 1 (spoofing / deepfake detection) evaluation partition of ASVspoof 5. | |
| Binary classification: bonafide vs. spoof. EER computed on the official | |
| ASVspoof5 track_1 eval protocol (680,774 utterances). | |
| evaluation_framework: inspect-ai | |
| tasks: | |
| - id: antispoofing_eval | |
| config: default | |
| split: test | |
| field_spec: | |
| input: audio | |
| target: label | |
| solvers: | |
| - name: speech_spoof_bench_solver | |
| scorers: | |
| - name: speech_spoof_scorer | |
| metrics: | |
| - eer_percent | |