NxAudio

NxAudio is an Elixir library for working with audio tensors, providing functionality similar to Python's torchaudio but built for the Nx ecosystem.

Errors related to the NxAudio framework itself.

Errors related to input/output operations.

Errors related to invalid data or inputs.

Errors that are unknown or unclassified

Default unexpected error

NX Implementation of common window functions.

Defines a behaviour for audio backend operations

Represents the metadata of an audio file

Defines how the backend should read the audio file.

Defines how the backend should save the audio tensor.

Implements a FFMPEG backend to deal with audio files. This module requires FFMPEG to be installed on the system.

Reading operations using the FFmpeg backend

Writing operations using the FFmpeg backend

Represents the supported audio encodings

Codec Names

Error when writing the tensor to a temporary file

Error raised when an audio file cannot be parsed using a backend tooling.

Error when a invalid configuration is provided to a backend tool.

Error when an invalid metadata is returned by a backend tool.

Represents the amplitude to decibel transformation for audio signals.

Error when a invalid configuration is provided to a transformation.

Implements Mel-scaled spectrograms - a perceptually-motivated time-frequency representation of audio.

Configuration options for mel spectrogram transformation.

Spectrogram transformation for audio tensors.

Configuration options for spectrogram transformation.

Error when a invalid configuration is provided to a visualization.

Provides a binned heatmap visualization for Nx-based Spectrograms.

Defines the configuration schema for the Spectrogram visualization.

Provides a waveform visualization for Nx-based audio tensors.

Defines the configuration schema for the Waveform visualization.