turian / audiomentations

A Python library for audio data augmentation. Inspired by albumentations. Useful for machine learning.

Audiomentations

A Python library for audio data augmentation. Inspired by albumentations. Useful for deep learning. Runs on CPU. Supports mono audio and partially multichannel audio. Can be integrated in training pipelines in e.g. Tensorflow/Keras or Pytorch. Has helped people get world-class results in Kaggle competitions. Is used by companies making next-generation audio products.

Need a Pytorch alternative with GPU support? Check out torch-audiomentations!

Setup

pip install audiomentations

Optional requirements

Some features have extra dependencies. Extra python package dependencies can be installed by running

pip install audiomentations[extras]

Feature	Extra dependencies
Load 24-bit wav files fast	wavio
LoudnessNormalization	pyloudnorm
Mp3Compression	ffmpeg and [pydub or lameenc]

Note: ffmpeg can be installed via e.g. conda or from the official ffmpeg download page.

Usage example

from audiomentations import Compose, AddGaussianNoise, TimeStretch, PitchShift, Shift
import numpy as np

SAMPLE_RATE = 16000

augment = Compose([
    AddGaussianNoise(min_amplitude=0.001, max_amplitude=0.015, p=0.5),
    TimeStretch(min_rate=0.8, max_rate=1.25, p=0.5),
    PitchShift(min_semitones=-4, max_semitones=4, p=0.5),
    Shift(min_fraction=-0.5, max_fraction=0.5, p=0.5),
])

# Generate 2 seconds of dummy audio for the sake of example
samples = np.random.uniform(low=-0.2, high=0.2, size=(32000,)).astype(np.float32)

# Augment/transform/perturb the audio data
augmented_samples = augment(samples=samples, sample_rate=SAMPLE_RATE)

Go to audiomentations/augmentations/transforms.py to see the waveform transforms you can apply, and what arguments they have.

See audiomentations/augmentations/spectrogram_transforms.py for spectrogram transforms.

Waveform transforms

Some of the following waveform transforms can be visualized (for understanding) by the audio-transformation-visualization GUI (made by phrasenmaeher), where you can upload your own input wav file

AddBackgroundNoise

Added in v0.9.0

Mix in another sound, e.g. a background noise. Useful if your original sound is clean and you want to simulate an environment where background noise is present.

Can also be used for mixup, as in https://arxiv.org/pdf/1710.09412.pdf

A folder of (background noise) sounds to be mixed in must be specified. These sounds should ideally be at least as long as the input sounds to be transformed. Otherwise, the background sound will be repeated, which may sound unnatural.

Note that the gain of the added noise is relative to the amount of signal in the input. This implies that if the input is completely silent, no noise will be added.

AddGaussianNoise

Added in v0.1.0

Add gaussian noise to the samples

AddGaussianSNR

Added in v0.7.0

Add gaussian noise to the samples with random Signal to Noise Ratio (SNR)

AddImpulseResponse

Added in v0.7.0

Convolve the audio with a random impulse response. Impulse responses can be created using e.g. http://tulrich.com/recording/ir_capture/

Some datasets of impulse responses are publicly available:

• EchoThief containing 115 impulse responses acquired in a wide range of locations.
• The MIT McDermott dataset containing 271 impulse responses acquired in everyday places.

Impulse responses are represented as wav files in the given ir_path.

AddShortNoises

Added in v0.9.0

Mix in various (bursts of overlapping) sounds with random pauses between. Useful if your original sound is clean and you want to simulate an environment where short noises sometimes occur.

A folder of (noise) sounds to be mixed in must be specified.

BandPassFilter

To be added in v0.18.0

Apply band-pass filtering to the input audio. The filter steepness is 6 dB per octave.

Clip

Added in v0.17.0

Clip audio by specified values. e.g. set a_min=-1.0 and a_max=1.0 to ensure that no samples in the audio exceed that extent. This can be relevant for avoiding integer overflow or underflow (which results in unintended wrap distortion that can sound horrible) when exporting to e.g. 16-bit PCM wav.

Another way of ensuring that all values stay between -1.0 and 1.0 is to apply PeakNormalization.

This transform is different from ClippingDistortion in that it takes fixed values for clipping instead of clipping a random percentile of the samples. Arguably, this transform is not very useful for data augmentation. Instead, think of it as a very cheap and harsh limiter (for samples that exceed the allotted extent) that can sometimes be useful at the end of a data augmentation pipeline.

ClippingDistortion

Added in v0.8.0

Distort signal by clipping a random percentage of points

The percentage of points that will be clipped is drawn from a uniform distribution between the two input parameters min_percentile_threshold and max_percentile_threshold. If for instance 30% is drawn, the samples are clipped if they're below the 15th or above the 85th percentile.

FrequencyMask

Added in v0.7.0

Mask some frequency band on the spectrogram. Inspired by https://arxiv.org/pdf/1904.08779.pdf

Gain

Added in v0.11.0

Multiply the audio by a random amplitude factor to reduce or increase the volume. This technique can help a model become somewhat invariant to the overall gain of the input audio.

Warning: This transform can return samples outside the [-1, 1] range, which may lead to clipping or wrap distortion, depending on what you do with the audio in a later stage. See also https://en.wikipedia.org/wiki/Clipping_(audio)#Digital_clipping

HighPassFilter

To be added in v0.18.0

Apply low-pass filtering to the input audio. The signal will be reduced by 6 dB per octave below the cutoff frequency, so this filter is fairly gentle.

LowPassFilter

To be added in v0.18.0

Apply low-pass filtering to the input audio. The signal will be reduced by 6 dB per octave above the cutoff frequency, so this filter is fairly gentle.

Mp3Compression

Added in v0.12.0

Compress the audio using an MP3 encoder to lower the audio quality. This may help machine learning models deal with compressed, low-quality audio.

This transform depends on either lameenc or pydub/ffmpeg.

Note that bitrates below 32 kbps are only supported for low sample rates (up to 24000 hz).

Note: When using the lameenc backend, the output may be slightly longer than the input due to the fact that the LAME encoder inserts some silence at the beginning of the audio.

LoudnessNormalization

Added in v0.14.0

Apply a constant amount of gain to match a specific loudness. This is an implementation of ITU-R BS.1770-4.

Warning: This transform can return samples outside the [-1, 1] range, which may lead to clipping or wrap distortion, depending on what you do with the audio in a later stage. See also https://en.wikipedia.org/wiki/Clipping_(audio)#Digital_clipping

Normalize

Added in v0.6.0

Apply a constant amount of gain, so that highest signal level present in the sound becomes 0 dBFS, i.e. the loudest level allowed if all samples must be between -1 and 1. Also known as peak normalization.

PitchShift

Added in v0.4.0

Pitch shift the sound up or down without changing the tempo

PolarityInversion

Added in v0.11.0

Flip the audio samples upside-down, reversing their polarity. In other words, multiply the waveform by -1, so negative values become positive, and vice versa. The result will sound the same compared to the original when played back in isolation. However, when mixed with other audio sources, the result may be different. This waveform inversion technique is sometimes used for audio cancellation or obtaining the difference between two waveforms. However, in the context of audio data augmentation, this transform can be useful when training phase-aware machine learning models.

Resample

Added in v0.8.0

Resample signal using librosa.core.resample

To do downsampling only set both minimum and maximum sampling rate lower than original sampling rate and vice versa to do upsampling only.

Reverse

To be added in v0.18.0

Reverse the audio. Also known as time inversion. Inversion of an audio track along its time axis relates to the random flip of an image, which is an augmentation technique that is widely used in the visual domain. This can be relevant in the context of audio classification. It was successfully applied in the paper AudioCLIP: Extending CLIP to Image, Text and Audio.

Shift

Added in v0.5.0

Shift the samples forwards or backwards, with or without rollover

TanhDistortion

To be added in v0.18.0

Apply tanh (hyperbolic tangent) distortion to the audio. This technique is sometimes used for adding distortion to guitar recordings. The tanh() function can give a rounded "soft clipping" kind of distortion, and the distortion amount is proportional to the loudness of the input and the pre-gain. Tanh is symmetric, so the positive and negative parts of the signal are squashed in the same way. This transform can be useful as data augmentation because it adds harmonics. In other words, it changes the timbre of the sound.

See this page for examples: http://gdsp.hf.ntnu.no/lessons/3/17/

TimeMask

Added in v0.7.0

Make a randomly chosen part of the audio silent. Inspired by https://arxiv.org/pdf/1904.08779.pdf

TimeStretch

Added in v0.2.0

Time stretch the signal without changing the pitch

Trim

Added in v0.7.0

Trim leading and trailing silence from an audio signal using librosa.effects.trim

Spectrogram transforms

SpecChannelShuffle

Added in v0.13.0

Shuffle the channels of a multichannel spectrogram. This can help combat positional bias.

SpecFrequencyMask

Added in v0.13.0

Mask a set of frequencies in a spectrogram, à la Google AI SpecAugment. This type of data augmentation has proved to make speech recognition models more robust.

The masked frequencies can be replaced with either the mean of the original values or a given constant (e.g. zero).

Known limitations

• Some transforms do not support multichannel audio yet. See Multichannel audio
• Expects the input dtype to be float32, and have values between -1 and 1.
• The code runs on CPU, not GPU. For a GPU-compatible version, check out pytorch-audiomentations
• Multiprocessing is not officially supported yet. See also #46

Contributions are welcome!

Multichannel audio

Most transforms, but not all, support 2D numpy arrays with shapes like (num_channels, num_samples)

The following table is valid for new versions of audiomentations, like >=0.17.0

Transform	Supports multichannel audio?
AddBackgroundNoise	No, 1D only
AddGaussianNoise	Yes
AddGaussianSNR	Yes
AddImpulseResponse	No, 1D only
AddShortNoises	No, 1D only
BandPassFilter	No, 1D only
Clip	Yes
ClippingDistortion	Yes
FrequencyMask	Yes
Gain	Yes
HighPassFilter	No, 1D only
LoudnessNormalization	Yes, up to 5 channels
LowPassFilter	No, 1D only
Mp3Compression	No, 1D only
Normalize	Yes
PitchShift	Yes
PolarityInversion	Yes
Resample	No, 1D only
Reverse	Yes
Shift	Yes
SpecChannelShuffle	Yes
SpecFrequencyMask	Yes
TanhDistortion	Yes
TimeMask	Yes
TimeStretch	Yes
Trim	No, 1D only

Changelog

Unreleased

Added

• Implement BandPassFilter, HighPassFilter, LowPassFilter, Reverse and TanhDistortion. Thanks to atamazian.

v0.17.0 (2021-06-25)

Added

• Add a fade option in Shift for eliminating unwanted clicks
• Add support for 32-bit int wav loading with scipy>=1.6
• Add support for float64 wav files. However, the use of this format is discouraged, since float32 is more than enough for audio in most cases.
• Implement Clip. Thanks to atamazian.
• Add some parameter sanity checks in AddGaussianNoise
• Officially support librosa 0.8.1

Changed

• Rename AddImpulseResponse to ApplyImpulseResponse. The former will still work for now, but give a warning.
• When looking for audio files in AddImpulseResponse, AddBackgroundNoise and AddShortNoises, follow symlinks by default.
• When using the new parameters min_snr_in_db and max_snr_in_db in AddGaussianSNR, SNRs will be picked uniformly in the decibel scale instead of in the linear amplitude ratio scale. The new behavior aligns more with human hearing, which is not linear.

Fixed

• Avoid division by zero in AddImpulseResponse when input is digital silence (all zeros)
• Fix inverse SNR characteristics in AddGaussianSNR. It will continue working as before unless you switch to the new parameters min_snr_in_db and max_snr_in_db. If you use the old parameters, you'll get a warning.

v0.16.0 (2021-02-11)

Added

• Implement SpecCompose for applying a pipeline of spectrogram transforms. Thanks to omerferhatt.

Fixed

• Fix a bug in SpecChannelShuffle where it did not support more than 3 audio channels. Thanks to omerferhatt.
• Limit scipy version range to >=1.0,<1.6 to avoid issues with loading 24-bit wav files. Support for scipy>=1.6 will be added later.

v0.15.0 (2020-12-10)

Added

• Add an option leave_length_unchanged to AddImpulseResponse

Fixed

• Fix picklability of instances of AddImpulseResponse, AddBackgroundNoise and AddShortNoises

v0.14.0 (2020-12-06)

Added

• Implement LoudnessNormalization
• Implement randomize_parameters in Compose. Thanks to SolomidHero.
• Add multichannel support to AddGaussianNoise, AddGaussianSNR, ClippingDistortion, FrequencyMask, PitchShift, Shift, TimeMask and TimeStretch

v0.13.0 (2020-11-10)

Added

• Lay the foundation for spectrogram transforms. Implement SpecChannelShuffle and SpecFrequencyMask.
• Configurable LRU cache for transforms that use external sound files. Thanks to alumae.
• Officially add multichannel support to Normalize

Changed

• Show a warning if a waveform had to be resampled after loading it. This is because resampling is slow. Ideally, files on disk should already have the desired sample rate.

Fixed

• Correctly find audio files with upper case filename extensions.
• Fix a bug where AddBackgroundNoise crashed when trying to add digital silence to an input. Thanks to juheeuu.

v0.12.1 (2020-09-28)

Changed

• Speed up AddBackgroundNoise, AddShortNoises and AddImpulseResponse by loading wav files with scipy or wavio instead of librosa.

v0.12.0 (2020-09-23)

Added

• Implement Mp3Compression
• Officially support multichannel audio in Gain and PolarityInversion
• Add m4a and opus to the list of recognized audio filename extensions

Changed

• Expand range of supported librosa versions

Removed

• Python <= 3.5 is no longer officially supported, since Python 3.5 has reached end-of-life
• Breaking change: Internal util functions are no longer exposed directly. If you were doing e.g. from audiomentations import calculate_rms, now you have to do from audiomentations.core.utils import calculate_rms

v0.11.0 (2020-08-27)

Added

• Implement Gain and PolarityInversion. Thanks to Spijkervet for the inspiration.

v0.10.1 (2020-07-27)

Changed

• Improve the performance of AddBackgroundNoise and AddShortNoises by optimizing the implementation of calculate_rms.

Fixed

• Improve compatibility of output files written by the demo script. Thanks to xwJohn.
• Fix division by zero bug in Normalize. Thanks to ZFTurbo.

v0.10.0 (2020-05-05)

Added

• AddImpulseResponse, AddBackgroundNoise and AddShortNoises now support aiff files in addition to flac, mp3, ogg and wav

Changed

• Breaking change: AddImpulseResponse, AddBackgroundNoise and AddShortNoises now include subfolders when searching for files. This is useful when your sound files are organized in subfolders.

Fixed

• Fix filter instability bug in FrequencyMask. Thanks to kvilouras.

v0.9.0 (2020-02-20)

Added

• Remember randomized/chosen effect parameters. This allows for freezing the parameters and applying the same effect to multiple sounds. Use transform.freeze_parameters() and transform.unfreeze_parameters() for this.
• Implement transform.serialize_parameters(). Useful for when you want to store metadata on how a sound was perturbed.
• Add a rollover parameter to Shift. This allows for introducing silence instead of a wrapped part of the sound.
• Add support for flac in AddImpulseResponse
• Implement AddBackgroundNoise transform. Useful for when you want to add background noise to all of your sound. You need to give it a folder of background noises to choose from.
• Implement AddShortNoises. Useful for when you want to add (bursts of) short noise sounds to your input audio.

Changed

• Disregard non-audio files when looking for impulse response files
• Switch to a faster convolve implementation. This makes AddImpulseResponse significantly faster.
• Expand supported range of librosa versions

Fixed

• Fix a bug in ClippingDistortion where the min_percentile_threshold was not respected as expected.
• Improve handling of empty input

v0.8.0 (2020-01-28)

Added

• Add shuffle parameter in Composer
• Add Resample transformation
• Add ClippingDistortion transformation
• Add fade parameter to TimeMask

Thanks to askskro

v0.7.0 (2020-01-14)

Added

• AddGaussianSNR
• AddImpulseResponse
• FrequencyMask
• TimeMask
• Trim

Thanks to karpnv

v0.6.0 (2019-05-27)

Added

• Implement peak normalization

v0.5.0 (2019-02-23)

Added

• Implement Shift transform

Changed

• Ensure p is within bounds

v0.4.0 (2019-02-19)

Added

• Implement PitchShift transform

Fixed

• Fix output dtype of AddGaussianNoise

v0.3.0 (2019-02-19)

Added

• Implement leave_length_unchanged in TimeStretch

v0.2.0 (2019-02-18)

Added

• Add TimeStretch transform
• Parametrize AddGaussianNoise

v0.1.0 (2019-02-15)

Added

• Initial release. Includes only one transform: AddGaussianNoise

Development

Install the dependencies specified in requirements.txt

Code style

Format the code with black

Run tests and measure code coverage

pytest

Generate demo sounds for empirical evaluation

python -m demo.demo

Alternatives

Audiomentations isn't the only python library that can do various types of audio data augmentation/degradation! Here's an overview:

Name	Github stars	License	Last commit	GPU support?
audio-degradation-toolbox
audio_degrader
audiomentations
AugLy
kapre
muda
nlpaug
pydiogment
python-audio-effects
sigment
SpecAugment
spec_augment
torch-audiomentations
torchaudio-augmentations
WavAugment

Acknowledgements

Thanks to Nomono for backing audiomentations.

Thanks to all contributors who help improving audiomentations.