# Audiomentations
[](https://circleci.com/gh/iver56/audiomentations)
[](https://codecov.io/gh/iver56/audiomentations)
[](https://github.com/ambv/black)
[](https://github.com/iver56/audiomentations/blob/main/LICENSE)
[](https://doi.org/10.5281/zenodo.13639627)
A Python library for audio data augmentation. Inspired by
[albumentations](https://github.com/albu/albumentations). Useful for deep learning. Runs on
CPU. Supports mono audio and 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-specific alternative with GPU support? Check out [torch-audiomentations](https://github.com/asteroid-team/torch-audiomentations)!
# Setup
[](https://pypi.org/project/audiomentations/)
[](https://pypi.org/project/audiomentations/)
`pip install audiomentations`
# Usage example
```python
from audiomentations import Compose, AddGaussianNoise, TimeStretch, PitchShift, Shift
import numpy as np
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(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=16000)
```
# Documentation
The API documentation, along with guides, example code, illustrations and example sounds, is available at [https://iver56.github.io/audiomentations/](https://iver56.github.io/audiomentations/)
# Transforms
* [AddBackgroundNoise](https://iver56.github.io/audiomentations/waveform_transforms/add_background_noise/): Mixes in another sound to add background noise
* [AddColorNoise](https://iver56.github.io/audiomentations/waveform_transforms/add_color_noise/): Adds noise with specific color
* [AddGaussianNoise](https://iver56.github.io/audiomentations/waveform_transforms/add_gaussian_noise/): Adds gaussian noise to the audio samples
* [AddGaussianSNR](https://iver56.github.io/audiomentations/waveform_transforms/add_gaussian_snr/): Injects gaussian noise using a randomly chosen signal-to-noise ratio
* [AddShortNoises](https://iver56.github.io/audiomentations/waveform_transforms/add_short_noises/): Mixes in various short noise sounds
* [AdjustDuration](https://iver56.github.io/audiomentations/waveform_transforms/adjust_duration/): Trims or pads the audio to fit a target duration
* [AirAbsorption](https://iver56.github.io/audiomentations/waveform_transforms/air_absorption/): Applies frequency-dependent attenuation simulating air absorption
* [Aliasing](https://iver56.github.io/audiomentations/waveform_transforms/aliasing/): Produces aliasing artifacts by downsampling without low-pass filtering and then upsampling
* [ApplyImpulseResponse](https://iver56.github.io/audiomentations/waveform_transforms/apply_impulse_response/): Convolves the audio with a randomly chosen impulse response
* [BandPassFilter](https://iver56.github.io/audiomentations/waveform_transforms/band_pass_filter/): Applies band-pass filtering within randomized parameters
* [BandStopFilter](https://iver56.github.io/audiomentations/waveform_transforms/band_stop_filter/): Applies band-stop (notch) filtering within randomized parameters
* [BitCrush](https://iver56.github.io/audiomentations/waveform_transforms/bit_crush/): Applies bit reduction without dithering
* [Clip](https://iver56.github.io/audiomentations/waveform_transforms/clip/): Clips audio samples to specified minimum and maximum values
* [ClippingDistortion](https://iver56.github.io/audiomentations/waveform_transforms/clipping_distortion/): Distorts the signal by clipping a random percentage of samples
* [Gain](https://iver56.github.io/audiomentations/waveform_transforms/gain/): Multiplies the audio by a random gain factor
* [GainTransition](https://iver56.github.io/audiomentations/waveform_transforms/gain_transition/): Gradually changes the gain over a random time span
* [HighPassFilter](https://iver56.github.io/audiomentations/waveform_transforms/high_pass_filter/): Applies high-pass filtering within randomized parameters
* [HighShelfFilter](https://iver56.github.io/audiomentations/waveform_transforms/high_shelf_filter/): Applies a high shelf filter with randomized parameters
* [Lambda](https://iver56.github.io/audiomentations/waveform_transforms/lambda/): Applies a user-defined transform
* [Limiter](https://iver56.github.io/audiomentations/waveform_transforms/limiter/): Applies dynamic range compression limiting the audio signal
* [LoudnessNormalization](https://iver56.github.io/audiomentations/waveform_transforms/loudness_normalization/): Applies gain to match a target loudness
* [LowPassFilter](https://iver56.github.io/audiomentations/waveform_transforms/low_pass_filter/): Applies low-pass filtering within randomized parameters
* [LowShelfFilter](https://iver56.github.io/audiomentations/waveform_transforms/low_shelf_filter/): Applies a low shelf filter with randomized parameters
* [Mp3Compression](https://iver56.github.io/audiomentations/waveform_transforms/mp3_compression/): Compresses the audio to lower the quality
* [Normalize](https://iver56.github.io/audiomentations/waveform_transforms/normalize/): Applies gain so that the highest signal level becomes 0 dBFS
* [Padding](https://iver56.github.io/audiomentations/waveform_transforms/padding/): Replaces a random part of the beginning or end with padding
* [PeakingFilter](https://iver56.github.io/audiomentations/waveform_transforms/peaking_filter/): Applies a peaking filter with randomized parameters
* [PitchShift](https://iver56.github.io/audiomentations/waveform_transforms/pitch_shift/): Shifts the pitch up or down without changing the tempo
* [PolarityInversion](https://iver56.github.io/audiomentations/waveform_transforms/polarity_inversion/): Flips the audio samples upside down, reversing their polarity
* [RepeatPart](https://iver56.github.io/audiomentations/waveform_transforms/repeat_part/): Repeats a subsection of the audio a number of times
* [Resample](https://iver56.github.io/audiomentations/waveform_transforms/resample/): Resamples the signal to a randomly chosen sampling rate
* [Reverse](https://iver56.github.io/audiomentations/waveform_transforms/reverse/): Reverses the audio along its time axis
* [RoomSimulator](https://iver56.github.io/audiomentations/waveform_transforms/room_simulator/): Simulates the effect of a room on an audio source
* [SevenBandParametricEQ](https://iver56.github.io/audiomentations/waveform_transforms/seven_band_parametric_eq/): Adjusts the volume of 7 frequency bands
* [Shift](https://iver56.github.io/audiomentations/waveform_transforms/shift/): Shifts the samples forwards or backwards
* [SpecChannelShuffle](https://iver56.github.io/audiomentations/spectrogram_transforms/): Shuffles channels in the spectrogram
* [SpecFrequencyMask](https://iver56.github.io/audiomentations/spectrogram_transforms/): Applies a frequency mask to the spectrogram
* [TanhDistortion](https://iver56.github.io/audiomentations/waveform_transforms/tanh_distortion/): Applies tanh distortion to distort the signal
* [TimeMask](https://iver56.github.io/audiomentations/waveform_transforms/time_mask/): Makes a random part of the audio silent
* [TimeStretch](https://iver56.github.io/audiomentations/waveform_transforms/time_stretch/): Changes the speed without changing the pitch
* [Trim](https://iver56.github.io/audiomentations/waveform_transforms/trim/): Trims leading and trailing silence from the audio
# Changelog
## [0.37.0] - 2024-09-03
### Changed
* Leverage the SIMD-accelerated [numpy-minmax](https://github.com/nomonosound/numpy-minmax) package for speed improvements. These transforms are faster now: `Limiter`, `Mp3Compression` and `Normalize`. Unfortunately, this removes support for Intel-based Macs. Intel Mac users have the following options: A) use audiomentations 0.36.1, B) Create a fork of audiomentations C) submit a patch to numpy-minmax, D) run Linux or Windows.
* Limit numpy dependency to >=1.21,<2 for now, since numpy v2 is not officially supported yet.
For the full changelog, including older versions, see [https://iver56.github.io/audiomentations/changelog/](https://iver56.github.io/audiomentations/changelog/)
# Acknowledgements
Thanks to [Nomono](https://nomono.co/) for backing audiomentations.
Thanks to [all contributors](https://github.com/iver56/audiomentations/graphs/contributors) who help improving audiomentations.
Raw data
{
"_id": null,
"home_page": "https://github.com/iver56/audiomentations",
"name": "audiomentations",
"maintainer": null,
"docs_url": null,
"requires_python": "<=3.12,>=3.8",
"maintainer_email": null,
"keywords": null,
"author": "Iver Jordal",
"author_email": null,
"download_url": "https://files.pythonhosted.org/packages/e3/9a/b88779750763ef49497200f2dbf48c2e9f852383eb8f01ef01543dd48643/audiomentations-0.37.0.tar.gz",
"platform": null,
"description": "# Audiomentations\r\n\r\n[](https://circleci.com/gh/iver56/audiomentations)\r\n[](https://codecov.io/gh/iver56/audiomentations)\r\n[](https://github.com/ambv/black)\r\n[](https://github.com/iver56/audiomentations/blob/main/LICENSE)\r\n[](https://doi.org/10.5281/zenodo.13639627)\r\n\r\nA Python library for audio data augmentation. Inspired by\r\n[albumentations](https://github.com/albu/albumentations). Useful for deep learning. Runs on\r\nCPU. Supports mono audio and multichannel audio. Can be\r\nintegrated in training pipelines in e.g. Tensorflow/Keras or Pytorch. Has helped people get\r\nworld-class results in Kaggle competitions. Is used by companies making next-generation audio\r\nproducts.\r\n\r\nNeed a Pytorch-specific alternative with GPU support? Check out [torch-audiomentations](https://github.com/asteroid-team/torch-audiomentations)!\r\n\r\n# Setup\r\n\r\n\r\n[](https://pypi.org/project/audiomentations/)\r\n[](https://pypi.org/project/audiomentations/)\r\n\r\n\r\n`pip install audiomentations`\r\n\r\n# Usage example\r\n\r\n```python\r\nfrom audiomentations import Compose, AddGaussianNoise, TimeStretch, PitchShift, Shift\r\nimport numpy as np\r\n\r\naugment = Compose([\r\n AddGaussianNoise(min_amplitude=0.001, max_amplitude=0.015, p=0.5),\r\n TimeStretch(min_rate=0.8, max_rate=1.25, p=0.5),\r\n PitchShift(min_semitones=-4, max_semitones=4, p=0.5),\r\n Shift(p=0.5),\r\n])\r\n\r\n# Generate 2 seconds of dummy audio for the sake of example\r\nsamples = np.random.uniform(low=-0.2, high=0.2, size=(32000,)).astype(np.float32)\r\n\r\n# Augment/transform/perturb the audio data\r\naugmented_samples = augment(samples=samples, sample_rate=16000)\r\n```\r\n\r\n# Documentation\r\n\r\nThe API documentation, along with guides, example code, illustrations and example sounds, is available at [https://iver56.github.io/audiomentations/](https://iver56.github.io/audiomentations/)\r\n\r\n# Transforms\r\n\r\n* [AddBackgroundNoise](https://iver56.github.io/audiomentations/waveform_transforms/add_background_noise/): Mixes in another sound to add background noise\r\n* [AddColorNoise](https://iver56.github.io/audiomentations/waveform_transforms/add_color_noise/): Adds noise with specific color\r\n* [AddGaussianNoise](https://iver56.github.io/audiomentations/waveform_transforms/add_gaussian_noise/): Adds gaussian noise to the audio samples\r\n* [AddGaussianSNR](https://iver56.github.io/audiomentations/waveform_transforms/add_gaussian_snr/): Injects gaussian noise using a randomly chosen signal-to-noise ratio\r\n* [AddShortNoises](https://iver56.github.io/audiomentations/waveform_transforms/add_short_noises/): Mixes in various short noise sounds\r\n* [AdjustDuration](https://iver56.github.io/audiomentations/waveform_transforms/adjust_duration/): Trims or pads the audio to fit a target duration\r\n* [AirAbsorption](https://iver56.github.io/audiomentations/waveform_transforms/air_absorption/): Applies frequency-dependent attenuation simulating air absorption\r\n* [Aliasing](https://iver56.github.io/audiomentations/waveform_transforms/aliasing/): Produces aliasing artifacts by downsampling without low-pass filtering and then upsampling\r\n* [ApplyImpulseResponse](https://iver56.github.io/audiomentations/waveform_transforms/apply_impulse_response/): Convolves the audio with a randomly chosen impulse response\r\n* [BandPassFilter](https://iver56.github.io/audiomentations/waveform_transforms/band_pass_filter/): Applies band-pass filtering within randomized parameters\r\n* [BandStopFilter](https://iver56.github.io/audiomentations/waveform_transforms/band_stop_filter/): Applies band-stop (notch) filtering within randomized parameters\r\n* [BitCrush](https://iver56.github.io/audiomentations/waveform_transforms/bit_crush/): Applies bit reduction without dithering\r\n* [Clip](https://iver56.github.io/audiomentations/waveform_transforms/clip/): Clips audio samples to specified minimum and maximum values\r\n* [ClippingDistortion](https://iver56.github.io/audiomentations/waveform_transforms/clipping_distortion/): Distorts the signal by clipping a random percentage of samples\r\n* [Gain](https://iver56.github.io/audiomentations/waveform_transforms/gain/): Multiplies the audio by a random gain factor\r\n* [GainTransition](https://iver56.github.io/audiomentations/waveform_transforms/gain_transition/): Gradually changes the gain over a random time span\r\n* [HighPassFilter](https://iver56.github.io/audiomentations/waveform_transforms/high_pass_filter/): Applies high-pass filtering within randomized parameters\r\n* [HighShelfFilter](https://iver56.github.io/audiomentations/waveform_transforms/high_shelf_filter/): Applies a high shelf filter with randomized parameters\r\n* [Lambda](https://iver56.github.io/audiomentations/waveform_transforms/lambda/): Applies a user-defined transform\r\n* [Limiter](https://iver56.github.io/audiomentations/waveform_transforms/limiter/): Applies dynamic range compression limiting the audio signal\r\n* [LoudnessNormalization](https://iver56.github.io/audiomentations/waveform_transforms/loudness_normalization/): Applies gain to match a target loudness\r\n* [LowPassFilter](https://iver56.github.io/audiomentations/waveform_transforms/low_pass_filter/): Applies low-pass filtering within randomized parameters\r\n* [LowShelfFilter](https://iver56.github.io/audiomentations/waveform_transforms/low_shelf_filter/): Applies a low shelf filter with randomized parameters\r\n* [Mp3Compression](https://iver56.github.io/audiomentations/waveform_transforms/mp3_compression/): Compresses the audio to lower the quality\r\n* [Normalize](https://iver56.github.io/audiomentations/waveform_transforms/normalize/): Applies gain so that the highest signal level becomes 0 dBFS\r\n* [Padding](https://iver56.github.io/audiomentations/waveform_transforms/padding/): Replaces a random part of the beginning or end with padding\r\n* [PeakingFilter](https://iver56.github.io/audiomentations/waveform_transforms/peaking_filter/): Applies a peaking filter with randomized parameters\r\n* [PitchShift](https://iver56.github.io/audiomentations/waveform_transforms/pitch_shift/): Shifts the pitch up or down without changing the tempo\r\n* [PolarityInversion](https://iver56.github.io/audiomentations/waveform_transforms/polarity_inversion/): Flips the audio samples upside down, reversing their polarity\r\n* [RepeatPart](https://iver56.github.io/audiomentations/waveform_transforms/repeat_part/): Repeats a subsection of the audio a number of times\r\n* [Resample](https://iver56.github.io/audiomentations/waveform_transforms/resample/): Resamples the signal to a randomly chosen sampling rate\r\n* [Reverse](https://iver56.github.io/audiomentations/waveform_transforms/reverse/): Reverses the audio along its time axis\r\n* [RoomSimulator](https://iver56.github.io/audiomentations/waveform_transforms/room_simulator/): Simulates the effect of a room on an audio source\r\n* [SevenBandParametricEQ](https://iver56.github.io/audiomentations/waveform_transforms/seven_band_parametric_eq/): Adjusts the volume of 7 frequency bands\r\n* [Shift](https://iver56.github.io/audiomentations/waveform_transforms/shift/): Shifts the samples forwards or backwards\r\n* [SpecChannelShuffle](https://iver56.github.io/audiomentations/spectrogram_transforms/): Shuffles channels in the spectrogram\r\n* [SpecFrequencyMask](https://iver56.github.io/audiomentations/spectrogram_transforms/): Applies a frequency mask to the spectrogram\r\n* [TanhDistortion](https://iver56.github.io/audiomentations/waveform_transforms/tanh_distortion/): Applies tanh distortion to distort the signal\r\n* [TimeMask](https://iver56.github.io/audiomentations/waveform_transforms/time_mask/): Makes a random part of the audio silent\r\n* [TimeStretch](https://iver56.github.io/audiomentations/waveform_transforms/time_stretch/): Changes the speed without changing the pitch\r\n* [Trim](https://iver56.github.io/audiomentations/waveform_transforms/trim/): Trims leading and trailing silence from the audio\r\n\r\n# Changelog\r\n\r\n## [0.37.0] - 2024-09-03\r\n\r\n### Changed\r\n\r\n* Leverage the SIMD-accelerated [numpy-minmax](https://github.com/nomonosound/numpy-minmax) package for speed improvements. These transforms are faster now: `Limiter`, `Mp3Compression` and `Normalize`. Unfortunately, this removes support for Intel-based Macs. Intel Mac users have the following options: A) use audiomentations 0.36.1, B) Create a fork of audiomentations C) submit a patch to numpy-minmax, D) run Linux or Windows.\r\n* Limit numpy dependency to >=1.21,<2 for now, since numpy v2 is not officially supported yet.\r\n\r\nFor the full changelog, including older versions, see [https://iver56.github.io/audiomentations/changelog/](https://iver56.github.io/audiomentations/changelog/)\r\n\r\n# Acknowledgements\r\n\r\nThanks to [Nomono](https://nomono.co/) for backing audiomentations.\r\n\r\nThanks to [all contributors](https://github.com/iver56/audiomentations/graphs/contributors) who help improving audiomentations.\r\n",
"bugtrack_url": null,
"license": "MIT",
"summary": "A Python library for audio data augmentation. Inspired by albumentations. Useful for machine learning.",
"version": "0.37.0",
"project_urls": {
"Changelog": "https://iver56.github.io/audiomentations/changelog/",
"Documentation": "https://iver56.github.io/audiomentations/",
"Homepage": "https://github.com/iver56/audiomentations",
"Issue Tracker": "https://github.com/iver56/audiomentations/issues"
},
"split_keywords": [],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "22a8d0833e7d02e71feb0e7c3e14b9857ff8faaa8137f97aaa422f007a108535",
"md5": "e2a28b90459b4390cafcfc41423c3726",
"sha256": "2c7b06f4268796fcaf9c66237fa3de994242b601d8a590928f679d12e211577f"
},
"downloads": -1,
"filename": "audiomentations-0.37.0-py3-none-any.whl",
"has_sig": false,
"md5_digest": "e2a28b90459b4390cafcfc41423c3726",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": "<=3.12,>=3.8",
"size": 80525,
"upload_time": "2024-09-03T07:35:15",
"upload_time_iso_8601": "2024-09-03T07:35:15.699173Z",
"url": "https://files.pythonhosted.org/packages/22/a8/d0833e7d02e71feb0e7c3e14b9857ff8faaa8137f97aaa422f007a108535/audiomentations-0.37.0-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "e39ab88779750763ef49497200f2dbf48c2e9f852383eb8f01ef01543dd48643",
"md5": "380de6d13632ca636763439dbff580b9",
"sha256": "cd8590cd5aed1c8f1c6bbfa17b2b7850e613e8a732436750b3385cdebb6d0c18"
},
"downloads": -1,
"filename": "audiomentations-0.37.0.tar.gz",
"has_sig": false,
"md5_digest": "380de6d13632ca636763439dbff580b9",
"packagetype": "sdist",
"python_version": "source",
"requires_python": "<=3.12,>=3.8",
"size": 79034,
"upload_time": "2024-09-03T07:35:18",
"upload_time_iso_8601": "2024-09-03T07:35:18.042003Z",
"url": "https://files.pythonhosted.org/packages/e3/9a/b88779750763ef49497200f2dbf48c2e9f852383eb8f01ef01543dd48643/audiomentations-0.37.0.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-09-03 07:35:18",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "iver56",
"github_project": "audiomentations",
"travis_ci": false,
"coveralls": true,
"github_actions": true,
"circle": true,
"requirements": [
{
"name": "audioread",
"specs": [
[
"==",
"2.1.9"
]
]
},
{
"name": "black",
"specs": []
},
{
"name": "coverage",
"specs": [
[
"==",
"7.4.4"
]
]
},
{
"name": "cylimiter",
"specs": [
[
"==",
"0.3.0"
]
]
},
{
"name": "fast-align-audio",
"specs": [
[
"==",
"0.3.0"
]
]
},
{
"name": "lameenc",
"specs": [
[
"==",
"1.4.2"
]
]
},
{
"name": "librosa",
"specs": [
[
"==",
"0.10.0.post2"
]
]
},
{
"name": "matplotlib",
"specs": [
[
"<",
"4"
],
[
">=",
"3.0.0"
]
]
},
{
"name": "numba",
"specs": [
[
"==",
"0.57.0"
]
]
},
{
"name": "numpy",
"specs": [
[
"==",
"1.23.0"
]
]
},
{
"name": "numpy-minmax",
"specs": [
[
">=",
"0.3.0"
],
[
"<",
"1"
]
]
},
{
"name": "numpy-rms",
"specs": [
[
"<",
"1"
],
[
">=",
"0.4.2"
]
]
},
{
"name": "pydub",
"specs": [
[
"==",
"0.23.1"
]
]
},
{
"name": "pyloudnorm",
"specs": [
[
"==",
"0.1.0"
]
]
},
{
"name": "pyroomacoustics",
"specs": [
[
"==",
"0.7.3"
]
]
},
{
"name": "pytest",
"specs": [
[
"==",
"7.4.4"
]
]
},
{
"name": "pytest-cov",
"specs": [
[
"==",
"5.0.0"
]
]
},
{
"name": "scipy",
"specs": [
[
"<",
"1.13"
],
[
">=",
"1.4"
]
]
},
{
"name": "soxr",
"specs": [
[
"==",
"0.3.5"
]
]
},
{
"name": "tqdm",
"specs": [
[
"==",
"4.66.3"
]
]
},
{
"name": "twine",
"specs": []
}
],
"lcname": "audiomentations"
}
JSON
