AlphaAvatar's picture
Migrate wav2vec2-large-robust-6L age/gender ONNX from plugins-persona branch speaker_attribute_onnx (byte-identical); correct licence to CC-BY-NC-SA-4.0 per verified audEERING provenance; add manifest, checksums, NOTICE and known-answer examples
3174195 verified
Raw
History Blame Contribute Delete
574 Bytes
{
"description": "The upstream model card's known-answer case: one second of digital silence at 16 kHz. Zero-mean/unit-variance normalisation leaves an all-zero signal unchanged, so this input is valid with or without it.",
"input_name": "signal",
"shape": [
1,
16000
],
"dtype": "float32",
"numpy_recipe": "np.zeros((1, 16000), dtype=np.float32)",
"input_sha256_of_raw_float32_bytes": "4f7988030a00d082fe445e00a2ac5dab502300ff1b80e8592dd569867b60ef74",
"first_8_values": [
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
]
}