auglib.transform

auglib.transform comes with a number of transforms, that can augment an audio signal.

import auglib
import numpy as np

sampling_rate = 8000
signal = np.ones((1, 4))
transform = auglib.transform.GainStage(-10)
transform(signal, sampling_rate)

array([[0.31622776, 0.31622776, 0.31622776, 0.31622776]], dtype=float32)

Transforms can be combined using auglib.transform.Compose.

transform = auglib.transform.Compose(
    [
        auglib.transform.HighPass(cutoff=3000),
        auglib.transform.Clip(),
        auglib.transform.NormalizeByPeak(peak_db=-3),
    ]
)

You can add your own Python based transform using auglib.transform.Function.

def repeat(signal, sampling_rate, repeats):
    import numpy as np
    return np.tile(signal, repeats)

transform = auglib.transform.Function(repeat, {'repeats': 3})

Base	Base class for transforms applied to a signal.
AMRNB	Encode-decode signal using AMRNB codec.
Append	Append an auxiliary signal.
AppendValue	Expand signal with a constant value.
BabbleNoise	Adds Babble Noise.
BandPass	Run signal through a band-pass filter.
BandStop	Run signal through a band-stop filter.
Clip	Hard/soft-clip the signal.
ClipByRatio	Hard/soft-clip a certain fraction of the signal.
Compose	Compose several transforms together.
CompressDynamicRange	Compress the dynamic range.
Fade	Fade-in and fade-out of signal.
FFTConvolve	Convolve signal with another signal.
Function	Apply a custom function to the signal.
GainStage	Scale signal by linear factor.
HighPass	Run audio signal through a high-pass filter.
LowPass	Run audio signal through a low-pass filter.
Mask	Masked transformation.
Mix	Mix two audio signals.
NormalizeByPeak	Peak-normalize the signal to a desired level.
PinkNoise	Adds pink noise.
Prepend	Prepend an auxiliary signal.
PrependValue	Prepend signal with a constant value.
Resample	Resample signal to another sampling rate.
Select	Randomly select from a pool of transforms.
Shift	Shift signal without changing its duration.
Tone	Adds basic waveform.
Trim	Trim, zero pad, and/or repeat signal.
WhiteNoiseGaussian	Adds Gaussian white noise.
WhiteNoiseUniform	Adds uniform white noise.