hsaim
/

ast-finetuned-gtzan

Audio Classification

audio-spectrogram-transformer

music-genre-classification

Eval Results (legacy)

Model card Files Files and versions

AST fine-tuned on GTZAN

This model fine-tunes MIT/ast-finetuned-audioset-10-10-0.4593 for 10-way music genre classification on GTZAN.

Evaluation

Validation accuracy: 91.00% on the seeded 90/10 GTZAN split.

Downloads last month: -

Safetensors

Model size

86.2M params

Tensor type

F32

·

Model tree for hsaim/ast-finetuned-gtzan

Base model

MIT/ast-finetuned-audioset-10-10-0.4593

Finetuned

(202)

this model

Dataset used to train hsaim/ast-finetuned-gtzan

Evaluation results

Accuracy on GTZAN
self-reported

0.910