Files
acousticbrainz-ng/models/genre_tzanetakis-musicnn-msd-2.json
2025-08-06 15:38:22 -04:00

80 lines
2.4 KiB
JSON

{
"name": "genre GTZAN",
"type": "multi-class classifier",
"link": "https://essentia.upf.edu/models/classifiers/genre_tzanetakis/genre_tzanetakis-musicnn-msd-2.pb",
"version": "1",
"description": "classification of music by genre",
"author": "Pablo Alonso",
"email": "pablo.alonso@upf.edu",
"release_date": "2020-07-07",
"framework": "tensorflow",
"framework_version": "1.15.0",
"classes": [
"blues",
"classical",
"country",
"disco",
"hiphop",
"jazz",
"metal",
"pop",
"reggae",
"rock"
],
"model_types": ["frozen_model"],
"dataset": {
"name": "the GTZAN Genre Collection",
"citation": "@article{tzanetakis2002musical,\n title={Musical genre classification of audio signals},\n author={Tzanetakis, George and Cook, Perry},\n journal={IEEE Transactions on speech and audio processing},\n volume={10},\n number={5},\n pages={293--302},\n year={2002},\n publisher={IEEE}\n}",
"size": "1000 track excerpts, 100 per genre",
"metrics": {
"5-fold_cross_validation_normalized_accuracy": 0.83
}
},
"schema": {
"inputs": [
{
"name": "model/Placeholder",
"type": "float",
"shape": [187, 96]
}
],
"outputs": [
{
"name": "model/Sigmoid",
"type": "float",
"shape": [1, 10],
"op": "Sigmoid",
"output_purpose": "predictions"
},
{
"name": "model/dense_2/BiasAdd",
"type": "float",
"shape": [1, 10],
"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"
}
}