pyAudioAnalysis Documentation

pyAudioAnalysis is a library for various audio analysis tasks, including:

• Feature Extraction: Extracting representations like mfccs, spectrogram, and chromagram.
• Classification: Training, tuning, and evaluating classifiers for audio segments and classifying unknown sounds.
• Detection & Segmentation: Detecting audio events, excluding silence, performing supervised segmentation (joint segmentation-classification), and unsupervised segmentation (e.g., speaker diarization).
• Regression: Training and using audio regression models (e.g., for emotion recognition).
• Visualization: Applying dimensionality reduction to visualize audio data and content similarities.

To install pyAudioAnalysis, clone the repository, install the required dependencies, and then install the package in editable mode using pip.

git clone https://github.com/tyiannak/pyAudioAnalysis.git
pip install -r ./requirements.txt
pip install -e .

You can use the audioTrainTest module (aliased as aT in the example) to train a classifier using folders of WAV files (where each folder name represents a class) and then classify an unknown audio file.

Note: The extract_features_and_train function requires windowing parameters like aT.shortTermWindow and aT.shortTermStep.

from pyAudioAnalysis import audioTrainTest as aT
aT.extract_features_and_train(["classifierData/music","classifierData/speech"], 1.0, 1.0, aT.shortTermWindow, aT.shortTermStep, "svm", "svmSMtemp", False)
aT.file_classification("data/doremi.wav", "svmSMtemp","svm")

pyAudioAnalysis provides command-line support for its functionalities. To extract the spectrogram of a WAV file, use the audioAnalysis.py script with the fileSpectrogram command and the -i flag for the input file.

python audioAnalysis.py fileSpectrogram -i data/doremi.wav