Skip to content

Feature Extraction

Feature extraction utilities for audio processing.

This module provides functions to extract features such as MFCCs and spectrograms.

Author: Esgr0bar

extract_mfcc(y, sr, n_mfcc=13)

Extract MFCC features from an audio signal.

Parameters:

Name Type Description Default
y ndarray

Audio time series.

 required
sr int

Sample rate of the audio.

 required
n_mfcc int

Number of MFCCs to return.

 13

Returns:

Type Description

numpy.ndarray: MFCC feature matrix.
Source code in src/feature_extraction.py 
12
13
14
15
16
17
18
19
20
21
22
23
24
25
def extract_mfcc(y, sr, n_mfcc=13):
    """
    Extract MFCC features from an audio signal.

    Args:
        y (numpy.ndarray): Audio time series.
        sr (int): Sample rate of the audio.
        n_mfcc (int): Number of MFCCs to return.

    Returns:
        numpy.ndarray: MFCC feature matrix.
    """
    mfccs = librosa.feature.mfcc(y=y, sr=sr, n_mfcc=n_mfcc)
    return mfccs

extract_spectrogram(y, sr, n_fft=2048, hop_length=512)

Extract spectrogram from an audio signal.

Parameters:

Name Type Description Default
y ndarray

Audio time series.

 required
sr int

Sample rate of the audio.

 required
n_fft int

Number of samples per FFT.

 2048
hop_length int

Number of samples between successive frames.

 512

Returns:

Type Description

numpy.ndarray: Spectrogram matrix.
Source code in src/feature_extraction.py 
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
def extract_spectrogram(y, sr, n_fft=2048, hop_length=512):
    """
    Extract spectrogram from an audio signal.

    Args:
        y (numpy.ndarray): Audio time series.
        sr (int): Sample rate of the audio.
        n_fft (int): Number of samples per FFT.
        hop_length (int): Number of samples between successive frames.

    Returns:
        numpy.ndarray: Spectrogram matrix.
    """
    S = np.abs(librosa.stft(y, n_fft=n_fft, hop_length=hop_length))
    return S