68 lines
1.9 KiB
JSON
68 lines
1.9 KiB
JSON
{
|
|
"name": "gender",
|
|
"type": "multi-class classifier",
|
|
"link": "https://essentia.upf.edu/models/classifiers/gender/gender-musicnn-msd-2.pb",
|
|
"version": "1",
|
|
"description": "classification of vocal music by gender (male/female)",
|
|
"author": "Pablo Alonso",
|
|
"email": "pablo.alonso@upf.edu",
|
|
"release_date": "2020-07-07",
|
|
"framework": "tensorflow",
|
|
"framework_version": "1.15.0",
|
|
"classes": ["female", "male"],
|
|
"model_types": ["frozen_model"],
|
|
"dataset": {
|
|
"name": "In-house MTG collection",
|
|
"size": "3311 full tracks, 1508/1803 per class",
|
|
"metrics": {
|
|
"5-fold_cross_validation_normalized_accuracy": 0.88
|
|
}
|
|
},
|
|
"schema": {
|
|
"inputs": [
|
|
{
|
|
"name": "model/Placeholder",
|
|
"type": "float",
|
|
"shape": [187, 96]
|
|
}
|
|
],
|
|
"outputs": [
|
|
{
|
|
"name": "model/Sigmoid",
|
|
"type": "float",
|
|
"shape": [1, 2],
|
|
"op": "Sigmoid",
|
|
"output_purpose": "predictions"
|
|
},
|
|
{
|
|
"name": "model/dense_2/BiasAdd",
|
|
"type": "float",
|
|
"shape": [1, 2],
|
|
"op": "fully connected",
|
|
"description": "logits",
|
|
"output_purpose": ""
|
|
},
|
|
{
|
|
"name": "model/dense_1/BiasAdd",
|
|
"type": "float",
|
|
"shape": [1, 100],
|
|
"op": "fully connected",
|
|
"description": "penultimate layer",
|
|
"output_purpose": ""
|
|
},
|
|
{
|
|
"name": "model/dense/BiasAdd",
|
|
"type": "float",
|
|
"shape": [1, 200],
|
|
"op": "fully connected",
|
|
"output_purpose": "embeddings"
|
|
}
|
|
]
|
|
},
|
|
"citation": "@inproceedings{alonso2020tensorflow,\n title={Tensorflow Audio Models in Essentia},\n author={Alonso-Jim{\\'e}nez, Pablo and Bogdanov, Dmitry and Pons, Jordi and Serra, Xavier},\n booktitle={IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},\n year={2020}\n}",
|
|
"inference": {
|
|
"sample_rate": 16000,
|
|
"algorithm": "TensorflowPredictMusiCNN"
|
|
}
|
|
}
|