metadata

datasets:
  - lewtun/music_genres_small
base_model:
  - facebook/wav2vec2-large
metrics:
  - accuracy
  - f1
tags:
  - audio
  - music
  - classification
  - Wav2Vec2
pipeline_tag: audio-classification

Music Genre Classification Model 🎶

This model classifies music genres based on audio signals. It was fine-tuned on the model Wav2Vec2 and using the datasets music_genres_small.

You can find a GitHub repository with an interface hosted by a Flask API to test the model: music-classifier repository

Metrics

Validation Accuracy: 75%
F1 Score: 74%
Validation Loss: 0.77

Example Usage

from transformers import Wav2Vec2ForSequenceClassification, Wav2Vec2FeatureExtractor
import librosa
import torch

# Genre mapping corrected to a dictionary
genre_mapping = {
    0: "Electronic",
    1: "Rock",
    2: "Punk",
    3: "Experimental",
    4: "Hip-Hop",
    5: "Folk",
    6: "Chiptune / Glitch",
    7: "Instrumental",
    8: "Pop",
    9: "International",
}

model = Wav2Vec2ForSequenceClassification.from_pretrained("gastonduault/music-classifier")
feature_extractor = Wav2Vec2FeatureExtractor.from_pretrained("facebook/wav2vec2-large")

# Function for preprocessing audio for prediction
def preprocess_audio(audio_path):
    audio_array, sampling_rate = librosa.load(audio_path, sr=16000)
    return feature_extractor(audio_array, sampling_rate=16000, return_tensors="pt", padding=True)

# Path to your audio file
audio_path = "./Nirvana - Come As You Are.wav"

# Preprocess audio
inputs = preprocess_audio(audio_path)

# Predict
with torch.no_grad():
    logits = model(**inputs).logits
    predicted_class = torch.argmax(logits, dim=-1).item()

# Output the result
print(f"song analized:{audio_path}")
print(f"Predicted genre: {genre_mapping[predicted_class]}")