Song Genre Visualisation and Classification

In the following venture we try to predict genres of songs by using a machine learning approach.

• The code aims to classify songs into different genres based on audio features such as tempo, spectral centroid, and rhythm.

• The dataset is a reduced version of the  GTZAN Dataset obtained from Kaggle. Here songs are organized by genre.

• The dataset is split into training and testing sets, and features are extracted from the audio files.

• Features include tempo, spectral centroid, and rhythm features obtained using librosa functions.

• The code uses a linear Support Vector Machine (SVM) classifier for genre classification.

• The SVM classifier is trained on the training set and evaluated on the testing set.

• Classification accuracy is calculated using scikit-learn's accuracy_score function.

Click here to access the dataset

20% of the dataset was used for testing, and the model had an accuracy of 50%

The following code is a tool to visualize audio signals in the time and frequency domains. It can plot the waveform, spectrogram and pitch contour for a given audio file.

A waveform is a graphical representation of a signal in the time domain, showing how the amplitude (strength) of the signal varies over time. 

A spectrogram is a visual representation of the spectrum of frequencies in a signal as they vary with time. It provides a way to analyze the frequency content of a signal over time, making it particularly useful for studying how the spectral characteristics of a sound change over different portions of an audio recording.

In a spectrogram, the horizontal axis represents time, the vertical axis represents frequency, and the color intensity (or darkness) represents the magnitude or power of the frequencies present at a particular time. 

A pitch contour is a graphical representation of how the pitch (frequency of a sound) varies over time in an audio signal. In the context of speech and music, pitch represents the perceived frequency of a sound, often associated with the perceived "highness" or "lowness" of the sound.

A pitch contour typically shows the changes in pitch over time, providing a visual representation of the pitch fluctuations in an audio signal.