Instructions to use laion/voiceclap-large-v2-genuineness with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use laion/voiceclap-large-v2-genuineness with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("audio-classification", model="laion/voiceclap-large-v2-genuineness")# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("laion/voiceclap-large-v2-genuineness", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Standalone VoiceCLAP-large-v2 genuineness predictor (bundled embedder + MLP head, val HTML)
04dc213 verified | """Minimal example: predict speech genuineness (0-6) for a wav file. | |
| Usage: | |
| python example.py path/to/clip.wav [full|balanced] | |
| The model argument is optional and defaults to ``full`` (best overall MAE/corr). | |
| Higher score = sounds more like a real, lived-in spoken moment; lower = | |
| more rehearsed / synthetic. | |
| """ | |
| import os | |
| import sys | |
| from genuineness_scorer import GenuinenessScorer | |
| def main(): | |
| if len(sys.argv) < 2: | |
| print("usage: python example.py <clip.wav> [full|balanced]") | |
| sys.exit(1) | |
| wav_path = sys.argv[1] | |
| model = sys.argv[2] if len(sys.argv) > 2 else "full" | |
| pkg_dir = os.path.dirname(os.path.abspath(__file__)) | |
| import torch | |
| device = "cuda" if torch.cuda.is_available() else "cpu" | |
| scorer = GenuinenessScorer(pkg_dir=pkg_dir, model=model, device=device) | |
| score = scorer.score(wav_path) | |
| print(f"[{model}] genuineness (0-6): {score:.3f}") | |
| if __name__ == "__main__": | |
| main() | |