---
metrics:
- accuracy
- roc_auc
---
Currently, the model returns the confidence score that the input audio is created by one of the following classical composers 
found in [MusicNet Dataset](https://www.kaggle.com/datasets/imsparsh/musicnet-dataset) - a curated collection of 330 freely-licensed 
labeled classical music recordings - used for the training of this model:
- [Beethoven](https://en.wikipedia.org/wiki/Ludwig_van_Beethoven);
- [Bach](https://en.wikipedia.org/wiki/Johann_Sebastian_Bach);
- [Schubert](https://en.wikipedia.org/wiki/Franz_Schubert);
- [Mozart](https://en.wikipedia.org/wiki/Wolfgang_Amadeus_Mozart);
- [Brahms](https://en.wikipedia.org/wiki/Johannes_Brahms).
  
More details in [my Kaggle notebook](https://www.kaggle.com/code/dima806/speech-accent-multiclass-classification) and [my Medium post](https://pub.towardsai.net/creating-an-audio-classification-pipeline-for-identifying-classical-music-composers-1b3a5c453a66?sk=804fec0f2a3ba90e69724f95e703d677).